Conventional approaches in Structural Health Monitoring (SHM) tend to be complex, destructive, and time-intensive. Additionally, they often require a large number of sensors to thoroughly assess structural integrity. In this study, we present a novel, non-destructive SHM framework based on machine learning (ML) for the accurate detection and localization of structural cracks. This approach leverages a minimal number of strain gauge sensors linked via Bluetooth communication. The framework is validated through empirical data collected from 3D carbon fiber-reinforced composites, including three distinct specimens, ranging from crack-free samples to specimens with up to ten cracks of varying lengths and depths. Strain data from five sensors were analyzed using a combination of Shewhart charts, Grubbs Test (GT), and a hierarchical clustering algorithm, specifically designed to evaluate and classify fractures. Our ML-based framework offers a streamlined and efficient alternative to traditional laboratory procedures, delivering precise crack detection with significant potential for applications in the composites industry.

Artificial Intelligence (AI) is fundamentally reshaping various industries by enhancing decision-making processes, optimizing operations, and unlocking new opportunities for innovation. This paper explores the applications of AI across four key sectors: healthcare, finance, manufacturing, and retail. Each section delves into the specific challenges faced by these industries, the AI technologies employed to address them, and the measurable impact on business outcomes and societal welfare. We also discuss the implications of AI integration, including ethical considerations, the future trajectory of AI development, and its potential to drive economic growth while posing challenges that need to be managed responsibly.

The objective of this research is to evaluate four distinct models for multi-population mortality projection in order to ascertain the most effective approach for forecasting the impact of COVID-19 pandemic on mortality. Utilising data from the Human Mortality Database for five countries Finland, Germany, Italy, the Netherlands, and the United States, the study identifies the generalized additive model (GAM) within the age-period-cohort (APC) analytical framework as the most promising for precise mortality forecasts. Consequently, this model serves as the basis for projecting the impact of the COVID-19 pandemic on future mortality rates. By examining various pandemic scenarios, ranging from mild to severe, the study concludes that projections assuming a diminishing impact of the pandemic over time are most consistent, especially for middle-aged and elderly populations. Projections derived from the superior GAM-APC model offer guidance for strategic planning and decision-making within sectors facing the challenges posed by extreme historical mortality events and the uncertain future mortality trajectories.

Effective communication in automated chat systems hinges on the ability to understand and respond to context. Traditional models often struggle with determining when additional context is necessary for generating appropriate responses. This paper introduces Context-Aware BERT (CA-BERT), a transformer-based model specifically fine-tuned to address this challenge. CA-BERT innovatively applies deep learning techniques to discern context necessity in multi-turn chat interactions, enhancing both the relevance and accuracy of responses. We describe the development of CA-BERT, which adapts the robust architecture of BERT with a novel training regimen focused on a specialized dataset of chat dialogues. The model is evaluated on its ability to classify context necessity, demonstrating superior performance over baseline BERT models in terms of accuracy and efficiency. Furthermore, CA-BERT's implementation showcases significant reductions in training time and resource usage, making it feasible for real-time applications. The results indicate that CA-BERT can effectively enhance the functionality of chatbots by providing a nuanced understanding of context, thereby improving user experience and interaction quality in automated systems. This study not only advances the field of NLP in chat applications but also provides a framework for future research into context-sensitive AI developments.

Occupational incidents in construction workplaces present a persistent challenge, with severe consequences for both workers and project outcomes. This study aims to examine the specific causal factors contributing to these incidents, focusing on the interaction between worker-related, environmental, and procedural factors. The objective is to identify key contributors to workplace accidents and develop a unified predictive model to mitigate future risks. To achieve this, we employed a machine learning approach on a comprehensive accidentology dataset collected from contractors across multiple construction sites. The dataset includes variables such as worker ex-perience, environmental conditions, adherence to safety protocols, and more. By analyzing direct, indirect, and root causes, the methodology uncovers hidden patterns and interdependencies that traditional analysis might overlook. The study’s findings indicate that worker experience and environmental factors are the most significant contributors to incident occurrence, with a clear interaction effect between these variables. The results not only confirm previous research but also offer enhanced predictive capabilities for future safety measures. This research demonstrates the value of machine learning in generating data-driven insights, ultimately aiding in the develop-ment of targeted interventions to improve safety standards in the construction industry.

Robotics offers innovative possibilities at all levels of education, which should be considered when planning the teaching process for certain topics of engineering mathematics in higher education. In this paper, we present a new teaching-learning technique using STEAM-based methods to explore the properties of cycloidal curves for Computer Science Engineering BSc students. Frontal teaching, which is part of the tradition of higher education, has been complemented with teaching methods that meet the generational needs of students: problem-based learning, integration of STEAM-based methods and project-based learning. We have developed a new methodological model that combines the advantages of the traditional teaching method with new forms of teaching that have proven to be effective in teaching Generation Z. The four-stage methodological model uses a variety of didactical approaches because each stage applies a different type of learning technique to make the learning process interesting, motivating and effective. One of the main pillars of the model is the use of Desmos dynamic geometry software to create animations, and the other is the use of educational robotics to support visualisation. The implementation of educational robotics and its potential for teaching cycloidal curves is discussed in detail. The effectiveness of the method was investigated with first-year undergraduate students in computer science engineering. Quantitative studies were carried out to test the effectiveness of the new teaching technique. The experimental results strongly indicate the value of using the four-stage methodological model to implement active learning in the field of cycloidal curves.

Background/Objectives: This review examines the evolution of lyotropic liquid crystals (LLCs) in ocular drug delivery, focusing on their ability to address the challenges associated with traditional ophthalmic formulations. The study aims to underscore the enhanced bioavailability, prolonged retention, and controlled release properties of LLCs that significantly improve therapeutic outcomes; Methods: The review synthesizes data from various studies on both bulk-forming LLCs and liquid crystal nanoparticles (LCNPs). It also considers advanced analytical techniques, including the use of machine learning and AI-driven predictive modeling, to forecast the phase behavior and molecular structuring of LLC systems. Emerging technologies in biosensing and real-time diagnostics are discussed to illustrate the broader applicability of LLCs in ocular health; Results: LLCs are identified as pivotal in promoting targeted drug delivery across different regions of the eye, with specific emphasis on the tailored optimization of LCNPs. The review highlights principal categories of LLCs used in ocular applications, each facilitating unique interactions with physiological systems to enhance drug efficacy and safety. Additionally, novel applications in biosensing demonstrate LLCs' capacity to improve diagnostic processes; Conclusions: Lyotropic liquid crystals offer transformative potential in ocular drug delivery by overcoming significant limitations of conventional delivery methods. The integration of predictive technologies and biosensing applications further enriches the utility of LLCs, indicating a promising future for their use in clinical settings. This review points to continued advancements and encourages further research in LLC technology to maximize its therapeutic benefits.

Power consumption prediction is a crucial component in enhancing the efficiency and sustainability of building operations. This study investigates the impact of data collection frequency and model selection on the predictive accuracy of power consumption in two distinct building types: an Academic one with 15-minute interval data and an Industrial one with hourly data. Various machine learning models, including Support Vector Machine (SVM) with Radial and Sigmoid kernels, Random Forest (RF), and Deep Neural Networks (DNN), across different data splits and feature sets. Our analysis reveals that higher data collection frequency generally improves model performance, as indicated by lower RMSE, MAPE, and CV values, alongside higher R² scores. The inclusion of more historical power consumption features has also been found to have a more significant impact on the accuracy of predictions than including climate condition features. Moreover, the SVM-Radial model consistently outperformed others, particularly in capturing complex, non-linear patterns in the data. However, the DNN model, while competent in some metrics, showed elevated MAPE values, suggesting potential overfitting issues. These findings suggest that careful consideration of data frequency, features and model selection is essential for optimizing power prediction, contributing to more efficient power management strategies in building operations.

The use of digital twin and shadow concepts for industrial material processes have promoted new notions to bridge the gap between physical and cyber manufacturing processes. Hence, many multi-disciplinary areas like advanced sensor technologies, material science, data analytics, and machine learning algorithms are employed to create such hybrid systems. On the other hand, new additive manufacturing (AM) processes for metals and polymers based on novel emerging technologies have already shown a promising horizon for manufacturing of sophisticated parts with complex geometries. These processes are going through a major transformation with the emergence of digital technology, hybrid physical-data driven modelling and fast reduced models. This paper presents a fresh look at the hybrid physical-data driven and reduced order modelling (ROM) techniques for digitalization of AM processes within a digital-twin concept. The main contribution is to show the benefits of ROM and machine learning (ML) technology for process data handling, optimization\controlling and their integration into the real-time assessment of AM processes. Hence, a novel combination of efficient data solver technology and architecturally designed neural network (NN) module has been developed for the transient manufacturing processes with high heating\cooling rates. Furthermore, a real-world case study has been presented where the combination of hybrid modelling along with ROM and ML schemes are used for an industrial wire arc AM (WAAM) process.

Dengue fever is a persistent public health issue in tropical regions, including Vietnam, where climate variability plays a crucial role in disease transmission dynamics. This study focuses on developing climate-based machine learning models to forecast dengue outbreaks in Ba Ria Vung Tau (BRVT) province, Vietnam, using meteorological data from 2003 to 2022. We utilized four predictive models—Negative Binomial Regression (NBR), Seasonal AutoRegressive Integrated Moving Average with Exogenous Regressors (SARIMAX), Extreme Gradient Boosting (XGBoost), and Long Short-Term Memory (LSTM)—to predict weekly dengue incidence. Key climate variables, including temperature, humidity, precipitation, and wind speed, were integrated into these models, with lagged variables included to capture delayed climatic effects on dengue transmission. The NBR model demonstrated the best performance in terms of predictive accuracy, achieving the lowest Mean Absolute Error (MAE) compared to other models. The inclusion of lagged climate variables significantly enhanced the model’s ability to predict dengue cases. SARIMAX and LSTM models, although effective in capturing seasonal trends, struggled with overfitting and failed to accurately predict short-term outbreaks. XGBoost exhibited moderate predictive power but was sensitive to overfitting, particularly without fine-tuning. Our findings confirm that climate-based machine learning models, particularly the NBR model, offer valuable tools for forecasting dengue outbreaks in BRVT. However, improving the models’ ability to predict short-term peaks remains a challenge. The integration of meteorological data into early warning systems is crucial for public health authorities to plan timely and effective interventions. This research contributes to the growing body of literature on climate-based disease forecasting and underscores the need for further model refinement to address the complexities of dengue transmission in highly endemic regions.

The learning system is primarily based on teaching, and teaching through digital resources is referred to as e-learning. As the world progresses towards success day by day, the use of technology, digital devices, and modern techniques is becoming increasingly prevalent. The global shift from traditional paperwork to digital learning is one of the most significant changes in the education sector. However, many factors have created numerous hurdles during the online setup. The Covid-19 pandemic has led to serious problems all over the world, with the education sector being one of the most affected. One of the greatest challenges students face during online education is the lack of digital infrastructure, especially in remote areas of Pakistan. The biggest challenge is connecting students with teachers due to limited internet access. Without proper internet connectivity, both students and teachers cannot access educational materials. In this context, the government plays a crucial role in providing adequate facilities to students, particularly in primary schools. If proper infrastructure is not made available during this pandemic, students will be deprived of a good quality education. Therefore, good infrastructure is essential for effective online education. With the proper resources in place, students will be able to receive a quality education.

Dengue fever remains one of the most significant public health challenges globally, particularly in tropical and subtropical regions. Traditional early warning systems (EWS) for dengue outbreaks, based on retrospective epidemiological data, suffer from delays and limited predictive accuracy. Modern systems, leveraging real-time data, artificial intelligence (AI), and machine learning (ML), improve accuracy but can be costly and difficult to implement in resource-constrained settings. This paper proposes a novel hybrid model that integrates the strengths of both traditional and modern EWS, with real-world examples from Brazil, and Malaysia. The hybrid model balances accuracy, scalability, and cost-effectiveness, making it a feasible option for low- and middle-income countries.

Plant diseases not only cause economic losses, but they also pose a health risk. In this regard, aflatoxins are linked to the development of some type of liver cancer, especially the hepatocellular carcinoma. Aflatoxins are usually produced by the fungi Aspergillus flavus and Parsiticus in figs, and their identification is usually carried out using invasive and complex methods or by visual inspection at an advanced stage. Given that Spain is one of the main producers of figs and Extremadura is the leading region in the country, it is necessary to explore alternative ways of detecting the presence of aflatoxins in figs and preventing them from entering the food chain. The aim of this research is to address the early detection of Aspergillus flavus fungus using non-invasive techniques with the aid of hyperspectral imaging and the application of artificial intelligence techniques, in particular deep learning. The images were taken after inoculation of the microtoxin using 3 different concentrations, related to three different classes and healthy figs (healthy controls). The analysis of the hyperspectral images was performed at the pixel level. Firstly, a full connected neural network was used to analyse the spectral signature associated with each pixel; second, the wavelet transform was applied to each spectral signature. The resulting images are fed to a convolutional neural network. The hyperparameters of the proposed models were adjusted based on the parameter tuning process also performed. The results are promising, with 83% accuracy, 82.75% recall and 83.25% F1-measure for the single neural network, and 77.25% accuracy, recall and F1 for the convolution neural network.

We present a data-driven, in situ proximal multi-sensor digital soil mapping approach to develop digital twins for multiple agricultural fields. A novel Digital Soil CoreTM (DSC) Probe was engineered that contains seven sensors, each of a distinct modality, including sleeve friction, tip force, dielectric permittivity, electrical resistivity, soil imagery, acoustics, and visible and near-infrared spectroscopy. The DSC System integrates components the DSC Probe, DSC software, and deployment equipment to sense soil characteristics at a high vertical spatial resolution (mm scale) along in situ soil profiles up to a depth of 120 cm in about 60 sec. The DSC Probe in situ proximal data are harmonized into a data cube providing vertical high-density knowledge associated with physical-chemical-biological soil conditions. In contrast, conventional ex situ soil samples derived from soil cores, soil pits, or surface samples analyzed using laboratory and other methods are bound by substantially coarser spatial resolution and multiple compounding errors. Our objective was to investigate the effects of mismatched scale between high-resolution in situ proximal sensor data and coarser resolution ex situ soil laboratory measurements to develop soil prediction models. Our study was conducted in central California soil in almond orchards. We collected DSC sensor data and spatially co-located soil cores that were sliced into narrow layers for laboratory-based soil measurements. Partial Least Squares Regression (PLSR) cross-validation was used to compare results testing four data integration methods. Method A reduced the high-resolution sensor data to discrete values paired with layer-based soil laboratory measurements. Method B used stochastic distributions of sensor data paired with layer-based soil laboratory measurements. Method C allocated the same soil analytical data to each one of the high-resolution multi-sensor data within a soil layer. Method D linked the high-density multi-sensor soil data directly to crop responses (crop performance and behavior metrics) bypassing costly laboratory soil analysis. Overall, the soil models derived from Method C outperformed Methods A and B. Soil predictions derived using Method D were most cost-effective for directly assessing soil-crop relationships, making this method well-suited for industrial-scale precision agriculture applications.

There is a need to find innovative learning methods that enable accelerated learning of ‎a foreign language. This study examined the effect of computer-assisted language ‎learning (CALL) in acquiring a foreign language, which combines cognitive and ‎emotional stimuli in the background. ‎ The study explored two factors related to the acquisition of a foreign language: the ‎duration and scope of the learning process and the depth of internalization of the newly ‎acquired language. Another objective was to assess the learning method in two ‎learning environments, a multimedia setting and a virtual reality headset (VRH), to ‎determine if the learning environment affects the learning results and leads to better ‎vocabulary retention. ‎ One hundred native French speakers, with an average age of 47.5, participated in the ‎study and had no prior knowledge of a new native language. We randomly divided the ‎participants into two groups (multimedia and VRH). They studied 550 words in a new ‎native language for five days: 30 minutes each evening and 15 minutes in the morning.‎ The post-learning test pointed out that both groups improved their vocabulary scores ‎significantly. Approximately one month after the learning experience, we administered ‎a knowledge retention test to 32 participants and found that the level of knowledge had ‎been retained. Finally, background variables (e.g., gender, age, previous knowledge of ‎the new native language) did not affect the learning results. ‎ The findings indicate that CALL, which integrates background cognitive and ‎emotional stimuli in both learning environments, significantly accelerates learning ‎pace, broadens the scope of newly acquired words, and ensures retention. The level of ‎improvement observed in our study is notably higher than reported in the literature ‎that had previously evaluated CALL and in-class language acquisition.

This study introduces the Data-Driven Personalized Learning Model (DDPLM), a sophisticated framework designed to enhance foreign language acquisition through the integration of big data analytics. Implemented within the educational platforms Edmodo and Duolingo, DDPLM utilizes real-time data processing to tailor learning paths and content dynamically to individual learner needs. Our findings indicate significant improvements in language learning efficiency, engagement, and retention. The model's adaptability across different digital environments showcases its potential scalability and effectiveness in various educational contexts. Additionally, the research addresses the critical role of personalized feedback and adaptive challenges in maintaining learner motivation and promoting deeper linguistic comprehension. The outcomes suggest that DDPLM significantly transforms traditional language education, making it more personalized, efficient, and aligned with individual learning preferences.

Monte Carlo techniques have been extensively used for planning laser-based clinical procedures such as photobiomodulation. However, the effect of several biological tissue characteristics, including morphological structure and physiological parameters, has not been carefully addressed in many applications. Specifically, many questions remain concerning the effect of skin phototype and body mass index on the efficacy of photobiomodulation for extraoral therapies. To address these questions, a Monte Carlo simulation model was developed to analyze the effects of body mass index-dependent skin structure and Fitzpatrick skin type, specifically tailored for the morphological characteristics of cheek tissue. The model describes the settings of a typical oral photobiomodulation treatment protocol for pain relief, namely the use of 660 nm and 808 nm laser wavelengths and a therapeutic dose of 2.0 J/cm² on the masseter muscle. The simulations were used to train a machine learning predictive model aimed at accelerating the treatment planning stage and assessing the importance of patient-specific parameters. A multiple regression approach was adopted to predict muscle dose and treatment time for the effective delivered dose.

Thanks to the rapid advancements in Connected and Automated Vehicles (CAVs) and vehicular communication, the concept of the Internet of Vehicles (IoV) combined with Artificial Intelligence (AI) and big data technologies promotes the vision of an Intelligent Transportation System (ITS). By enabling a comprehensive data exchange platform, ITS is critical in enhancing road safety, traffic efficiency, and the overall driving experience. However, the open and dynamic nature of IoV networks brings significant performance and security challenges to IoV data acquisition, storage, and usage. To comprehensively tackle these challenges, this paper proposes a Decentralized Digital Watermarking framework for smart Vehicular networks (D2WaVe). Specifically, D2WaVe consists of two core components: FIAE-GAN, a novel feature-integrated and attention-enhanced robust image watermarking model based on a Generative Adversarial Network (GAN), and BloVA, a Blockchain-based Video frames Authentication scheme. By leveraging an encoder-noise-decoder framework, trained FIAE-GAN watermarking models can achieve the invisibility and robustness of watermarks that can be embedded in video frames to verify the authenticity of video data. Then, BloVA ensures the integrity and auditability of IoV data in the storing and sharing stages. Experimental results based on a proof-of-concept prototype implementation validate the feasibility and effectiveness of our D2WaVe scheme for securing and auditing video data exchange in smart vehicular networks.

Background: Synovial sarcoma is a rare soft tissue tumor that constitutes 5-10% of all soft tissue sarcomas. Early diagnosis and risk stratification are essential for effective management. This study leverages machine learning models to improve the risk stratification of mortality in synovial sar-coma patients. Methods: This retrospective cohort study utilized data from the Surveillance, Epidemiology, and End Results (SEER) database to analyze patients diagnosed with synovial sarcoma between 2004 and 2015, as well as a validation cohort diagnosed from 2018 onward. The dataset encompassed demographic data, clinical characteristics, staging, treatment modalities, and outcomes. Four ma-chine learning models—support vector classifier (SVC), k-nearest neighbors (KNN), Gaussian naive Bayes, and gradient boosting—were trained and evaluated. Model performance was assessed using sensitivity, specificity, AUC-ROC, and Brier score. SHAP analysis was performed to determine the most influential features impacting model predictions. The top-performing model was validated using the 2018+ cohort. Results: The study included a total of 762 patients, with an average age of 39.4 years. The support vector classifier (SVC) outperformed the other models, achieving an AUC-ROC of 0.8153 (95% CI: 0.7578 to 0.8677) and a Brier score of 0.1715. Upon external validation with the 2018+ cohort, the SVC model yielded an AUC-ROC of 0.8179 (95% CI: 0.7632 to 0.8666). Key prognostic factors identified through SHAP analysis included tumor size, patient age, presence of metastasis, tumor differentiation, and cancer stage. Conclusion: Machine learning models can be used to stratify the risk of death in synovial sarcoma patients effectively, with the support vector classifier showing the most promise.

This study investigated the effectiveness of virtual reality (VR) technology in table tennis education compared to traditional training methods. A 12-week randomized controlled trial was conducted with 120 participants divided equally between VR and traditional training groups. Performance metrics, learning motivation, and satisfaction were assessed at regular intervals. Results demonstrated significant advantages of VR training, with the VR group showing superior improvements in serve accuracy (23.5% vs. 15.8%, p < 0.001), rally endurance (increase of 8.2 vs. 5.7 shots, p < 0.01), and overall skill scores (18.7 vs. 13.2 points improvement, p < 0.001). The VR group also exhibited higher increases in learning motivation (23.5% vs. 12.8%, p < 0.001) and satisfaction (31.5% vs. 18.7%, p < 0.001). Subgroup analysis revealed particular benefits for novice players and younger participants. These findings suggest that VR technology offers a promising approach to enhance table tennis education, potentially revolutionizing sports training methodologies. Future research should focus on long-term skill retention and the optimization of VR training protocols.

Education 4.0 arises to provide citizens with the technical/digital competencies and cognitive/interpersonal skills demanded by Industry 4.0. New technologies drive this change, though time-independent learning remains a challenge, because students might face a lack of support, advice and surveillance when teachers are unavailable. This study proposes complementing presential lessons with online learning driven by ChatGPT, applied as an educational tool able to mentor K-12 students learning science at home. First, ChatGPT’s performance in the field of K-12 science is evaluated, scoring A (9.3/10 in 2023, and 9.7/10 in 2024) and providing detailed, analytic, meaningful, and human-like answers. Then, an empirical interventional study is performed to assess the impact of using ChatGPT as a virtual mentor on real K-12 students. After the intervention, the grades of students in the experimental group improved 30%, and 70% of students stated a positive perception of the AI, suggesting a positive impact of the proposed educational approach. After discussion, the study concludes ChatGPT might be a useful educational tool able to provide K-12 students learning science with the functional and social/emotional support they might require, democratizing a higher level of knowledge acquisition and promoting students’ autonomy, security and self-efficacy. The results probe ChatGPT’s remarkable capacity (and immense potential) to assist teachers in their mentoring tasks, laying the foundations of virtual mentoring and paving the way for future research aimed at obtaining a more realistic view of the AI impact on education.

As telecommunication networks continue to grow, the volume, variety, and velocity of such `big data' poses an immense challenge for security and service analysts. A primary data format in such networks are Call Detail Records (CDRs), and more recently, eXtended Detail Records (XDRs), that capture a variety of information such as caller, receiver, duration, carrier, datetime, amongst other attributes. In this short paper, we investigate the current research trends and challenges that exist within this data-rich domain. As part of this, we look at the domain-specific tasks that analysts will seek to address, we look at how machine learning has been utilised to date within this domain, and we present our current findings on open research problems that are ripe for further investigation.

Detecting people and technical objects in various situations, such as natural disasters and warfare, is critical to search and rescue operations and the safety of the civilian. A fast and accurate detection for people and equipment can significantly increase the effectiveness of search and rescue missions and provide timely assistance to people. Computer vision and deep learning technologies play a key role in detecting the required objects due to their ability to analyse big volumes of visual data in real time. The performance of the neural networks such as YOLOv4-v8, Faster R-CNN, SSD, and EfficientDet has been analysed using COCO2017, SARD, SeaDronesSee, and VisDrone2019 datasets. Main criteria for comparison were mAP, Precision, Recall, F1-Score, and the ability of the neural network to work in real time. The most important metrics for evaluating the efficiency and performance of models for a given task are accuracy (mAP), F1-Score, and processing speed (FPS). These metrics allow us to evaluate both the accuracy of object recognition and the ability to use the models in real-world environments where high processing speed is important. Although different neural networks perform better on certain types of metrics, YOLO outperforms them on all metrics, showing the best results of mAP-0.88, F1 - 0.88, and FPS - 48, so the focus was on these models.

Background: Over the past few years there has been immense evolution in various areas particularly in the areas of digital technologies wherein the pace of change is very high. Industrial areas such as operations and supply chain management together with advanced technologies such as machine learning, big data analytics, artificial intelligence, as well as the Internet of Things, create completely different forms of operational models for various industries. In the area of healthcare too, these emerging computational sophistication is introduced to revolutionise the approaches to prevent, diagnose and treat diverse diseases and illnesses. Objective: The objective of this study is to provide an extensive review of the contemporary approaches utilizing data to cope with significant mental disorders. From over 60 relevant scholarly articles published between 2011 and 2023, it discusses how tools such as predictive modelling, social media analysis, data from smartphones, and chatbots help with issues such as early detection, telemonitoring, provision of psychological support, and individualised prevention. Method: An initial literature review to analyse over 60 research articles, which include empirical studies that were conducted between 2011 and 2023. The research assessed implemented novel digital approaches to mental health interventions including big data analytics for predicting condition status, machine learning for examining social media content, behaviour monitoring through smartphone sensors, and using conversational agents or chatbots. The following is an overview of general conclusions from experimental and descriptive secondary research studies published in professional outlets concerning possible advantages and disadvantages of data science applied to important concerns in mental health. Results: Research reveals that integrating subtle e-health tools in tandem with typical treatment approaches holds the potential to expand mental health services to more or less integrate them into clients’ day-to-day lives, and practically individualize effective treatments accordingly. Technological solutions for instance allow remote risk assessment, symptom monitoring and determination of treatment compliance. New lines of virtualized paradigm solve social challenges that interfere with the conventional provision and consumption of care. However, questions of privacy and the long-term effects as well as clinical adoption are yet to be solved in a analytically distinct manner.

Childhood is an obvious period for motor learning since children’s musculoskeletal and nervous systems are still in development. Adults adapt movements based on reward feedback about success and failure but it is less established whether school-age children also exhibit such reward-based motor learning. We designed a new ‘circle-drawing’ task suitable for assessing reward-based motor learning in both children and adults. Participants drew circles with their unseen hand on a tablet. They received binary reward feedback after each attempt based on the proximity of the average radius of their drawing to a target radius set as double the radius of their baseline drawings. We rewarded about 50% of the trials based on a performance-dependent reward criterion. Both children (7-17 years old) and adults increased the radius of their drawings in the direction of the target radius. We observed no difference in learning between children and adults. Moreover, both groups changed the radius less following reward than following reward absence, which is a sign of reward-based motor learning. We conclude that school-age children, like adults, exhibit reward-based motor learning.

This research presents an ensemble model developed to enhance the safety of bedridden patients in healthcare settings. The model utilizes a unique dataset capturing six typical movements performed by patients laying in bed, such as rolling over, falling off the bed, etc. The aim is to automate the classification of these movements, thereby enabling real-time monitoring of patients without the constant presence of hospital personnel in the room, a common practice in current healthcare settings. The dataset is preprocessed and balanced using the Synthetic Minority Over-sampling Technique, and then split into training and test sets for model training and evaluation. Three different models are explored: a Decision Tree Regressor, a Gradient Boosting Regressor, and a Bagging Regressor. The Bagging Regressor is an ensemble model that utilizes the Decision Tree Regressor as its base regressor. Each model’s performance is evaluated using accuracy, R2 score, and Mean Squared Error on both the training and test sets. The ensemble model demonstrates promising results, indicating the potential of machine learning to improve patient safety in healthcare settings. The study also employs K-Fold cross-validation and learning curves to validate the robustness of the ensemble model. This research contributes to the growing body of knowledge on the application of machine learning in healthcare and opens up new avenues for improving patient safety through real-time monitoring and classification of patient movements.

Attention mechanisms have become a fundamental component of deep learning, including the field of computer vision. The key idea behind attention in computer vision is to help the model focus on the relevant spatial regions of the input image, rather than treating all regions equally. The traditional approaches to attention mechanisms in computer vision often suffer from distribution inconsistencies in the attention maps, resulting in sharp transitions that negatively affect model’s focus and lead to poor generalization on complex shapes. The problem of spatial incoherence is particularly pronounced in the task of semantic segmentation, where accurate pixel-level predictions require a detailed understanding of the spatial relationships within the image. In this paper, we propose an attention mechanism called Smooth Attention designed for convolutional neural networks to address the problem of spatial inconsistency in attention maps through multidimensional spatial smoothing. We conduct a series of experiments to evaluate the effectiveness of the proposed mechanism and demonstrate its superior performance compared to traditional methods.

This research explores how regional socioeconomic variables affect the perception of social trust and support networks (PYCC) in Italian regions, and examines the implications for public policy designed to strengthen social cohesion. This study examines the variable "People You Can Count On" (PYCC) from the ISTAT-BES dataset, focusing on its distribution across Italian regions between 2013 and 2022. Using clustering through a k-Means algorithm optimized with the Silhouette coefficient and the Elbow method, three distinct clusters of regions emerged, highlighting significant differences in social support networks. An econometric model was employed to estimate the PYCC variable, factoring in socioeconomic indicators such as employment rates, income inequality, and social participation. The results indicate a complex interplay between socioeconomic conditions and social trust, with regions in the South and Islands showing increased community support, while many Northern regions experienced declines. The study suggests that areas with lower economic conditions often foster stronger social networks, driven by necessity. These findings underline the importance of targeted public policies aimed at fostering social cohesion, particularly in regions facing economic challenges. Policy implications include enhancing education, supporting small enterprises, and promoting social housing and welfare initiatives. Strengthening community participation and volunteering are also highlighted as critical strategies to build resilient support networks. Overall, the research provides valuable insights into the regional disparities of social trust and the role of socioeconomic factors in shaping community support across Italy.

Deep-learning-based convection schemes receive wide attention due to its impressive im-provement on precipitation distribution and tropical convections of earth system simulation. But they cannot represent the stochasticity of moist physics, which will degrade the simulation of large-scale circulations, climate mean, and variability. To solve this problem, a stochastic pa-rameterization scheme based on probabilistic diffusion model named DIFF-MP is developed. The cloud-resolving data from GRIST model is coarse-grained into resolved-scale variables and sub-grid contributions due to moist physics to form the training data. DIFF-MP’s performance is compared against generative adversarial network and variational autoencoder. Results show that DIFF-MP is consistently better than the other two models on prediction error, coverage ratio, and spread-skill correlation. The standard deviation, skewness, and kurtosis of subgrid contributions generated by DIFF-MP is also closer to the testing data than the others. Interpretability experiment shows that DIFF-MP’s parameterization of moist physics is physically reasonable.

Background: Although confirmed cases of infection with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) have been declining since late 2020 due to general vaccination, little research has been done regarding the impact of vaccines against SARS-CoV-2 in Spain in terms of hospitalizations and deaths. Objective: Our aim was to identify the reduction in severity and mortality of coronavirus disease 2019 (COVID-19) at a nationwide level due to vaccination. Methods: We designed a retrospective, population-based study to define waves of infection and to describe the characteristics of the hospitalized population. We also studied the rollout of vaccination and its relationship with the decline in hospitalizations and deaths. Finally, we developed two mathematical models to estimate non-vaccination scenarios using machine learning modeling (with the ElasticNet and RandomForest algorithms). The vaccination and non-vaccination scenarios were eventually compared to estimate the number of averted hospitalizations and deaths. Results: In total, 498,789 patients were included, with a global mortality of 14.3%. We identified six waves or epidemic outbreaks during the observed period. We established a strong relationship between the beginning of vaccination and the decline in both hospitalizations and deaths due to COVID-19 in all age groups. We also estimated that vaccination prevented 170,959 hospitalizations (CI 95% 77,844–264,075) and 24,546 deaths (CI 95% 2,548–46,543) in Spain between March 2021 and December 2021. We estimated a global reduction of 9.19% in total deaths during the first year of COVID-19 vaccination. Conclusions: Demographic and clinical profiles changed over the first months of the pandemic. Patients over 80 years old and other age groups obtained clinical benefit from early vaccination. The severity of COVID-19, in terms of hospitalizations and deaths, decreased due to vaccination.

Background: Atherosclerosis is a leading cause of cardiovascular disease, characterized by lipid accumulation and chronic inflammation within arterial walls. Early detection in young adults is crucial for preventing adverse cardiovascular events. This study investigates the associations between inflammatory indices, lipid biomarkers, and the presence of atherosclerosis in patients aged 18 to 55 years. Methods: A cross-sectional study was conducted involving 89 participants divided into two groups: 62 patients with documented atherosclerosis (main group) and 27 healthy controls without significant atherosclerosis. Comprehensive data—including demographic information, medication use, imaging results, laboratory parameters, and calculated inflammatory indices (SIRI, SII, AISI, NLR, PLR, MLR)—were collected. Statistical analyses included correlation assessments, group comparisons using the Mann-Whitney U test, logistic regression modeling, feature importance analysis with Random Forest and Gradient Boosting classifiers, receiver operating characteristic (ROC) curves, and K-means clustering. Results: Significant differences were observed between the main and control groups. Patients with atherosclerosis exhibited elevated inflammatory indices (SIRI, NLR, MLR, SII) and lipid profile abnormalities (higher TC and LDL-C, lower HDL-C). Lp(a) and ANGPTL3 levels were significantly higher in the main group (p

Background: Bacterial/fungal coinfections (BFC) are associated with antibiotic overuse, poor outcomes such as prolonged ICU stay and increased mortality. Our aim was to develop machine learning-based predictive models to identify respiratory bacterial or fungal coinfections on ICU admission. Methods: Secondary analysis of two prospective multicenter cohort studies with confirmed influenza A(H1N1)pdm09 and COVID-19. Multiple logistic regression (MLR) and random forest (RF) were used to identify factors associated with BFC in the overall population and in each subgroup (influenza and COVID-19). Performance was assessed by area under the ROC curve (AUC) and out-of-bag (OOB) for MLR and RF, respectively. Results: Of the 8,902 patients, 41.6% had influenza and 58.4% had SARS-CoV-2 infection. The median age was 60 years, 66% were male and the crude ICU mortality was 25%. BFC was observed in 14.2% of patients. Overall predictive models showed modest performance with AUC of 0.68 (MLR) and OOB 36.9% (RF). Specific models did not show improved performance. However, age, procalcitonin, CRP, APACHE II, SOFA and shock were factors associated with BFC in most models. Conclusion: Machine learning models for predicting respiratory BFC in patients with pandemic viruses are not able to predict adequately. However, the presence of factors such as advanced age, elevated procalcitonin or CPR and high severity of illness should alert clinicians to the need to rule out this complication on admission to the ICU.

Imitation learning, also known as programming by demonstration, has been shown to be a promising paradigm for intuitive robot programming by non-expert users. However, the classical kinesthetic approach with physical hand guidance suffers from generalizability across different robot types and is impractical for demonstrating tasks with long horizons. Visual imitation learning enables the recording of multi-step tasks as a single continuous video, allowing non-experts to demonstrate tasks naturally. Existing approaches typically require a large amount of data to develop end-to-end deep learning models that map raw pixels to robot actions. This paper explores the application of visual imitation learning from one-shot demonstration, significantly reducing the data requirements and simplifying the programming process. To achieve this target, a framework is proposed to map hand trajectories to the robot end-effector, consisting of four essential components: hand detection; object detection; segmentation of the trajectories into elemental skills; and learning the skills. Methods are developed for each component and evaluated on recorded videos to demonstrate the effectiveness of the proposed framework.

The success of deep learning (DL) in offline brain-computer interfaces (BCIs) has not yet translated into efficient online applications. This is due to three primary challenges. First, most DL solutions are designed for offline decoding, making the transition to online decoding complex and unclear. Second, the use of sliding windows in online decoding increases the computational complexity of DL training. Third, DL models typically require large amounts of training data, which are often scarce in BCI applications. To address these issues and enable real-time decoding, even across different subjects without calibration data, we first introduce a novel approach called real-time adaptive pooling (RAP). RAP modifies the pooling layers of existing offline DL models towards the online decoding requirements. Additionally, it significantly reduces the computational demand during training by jointly decoding consecutive windows. To reduce the amount of training data required, our approach leverages different levels of domain adaptation. We show how different settings enable different adaptation solutions. The results demonstrate that our approach is both powerful and can be calibration-free, providing a robust and practical solution for real-time BCI applications. These findings pave the way for the development of co-adaptive and highly efficient DL-based online BCI systems.

The growing use of drones in precision agriculture highlights the need for enhanced operational efficiency. Despite the ability of computer vision based on deep learning has made remarkable progress in the past ten years, when it comes to segmentation task on UAVs, there is always a conflict between the demand of high precision and low inference latency. Due to such a dilemma, we propose the IRB-YOLO, an efficiency model based on Inverted Residual Block, devoting to provide constructive strategies in real-time detection tasks of UAV camera. The working details of this paper are as follows: (1) This paper innovates with a IR-Block(Inverted Residual Block), integrated into a refined YOLOv8-seg structure to create IRB-YOLO. This model specializes in pixel-level classification of UAV-acquired RGB images, facilitating the creation of exact maps to guide agricultural strategies. (2)When it comes to the experiments on a Vatica dataset with any other light-weight segmentation model, IRB-YOLO achieve at least a 3.3% increase in mAP. Further validation using a diverse species dataset confirms its robust generalization. (3)Without overloading the complex attention mechanism and deeper and deeper network, a stem that incorporates efficient feature extraction components, inverted residual block, can still possess outstanding modeling capabilities. IRB-YOLO builds a bridge between academic research and edge deployment of drones, making it applicable in real-world scenarios.

The multi-UAV target search problem is crucial in the field of autonomous Unmanned Aerial Vehicle (UAV) decision-making. The algorithm design of Multi-agent Reinforcement Learning (MARL) methods has become integral to research on multi-UAV target search, due to its adaptability to the rapid online decision-making required by UAVs in complex, uncertain environments. In non-cooperative target search scenarios, targets may have the ability to escape, complicating UAVs' search efforts and hindering the convergence of the MARL algorithm training. This paper investigates the multi-UAV target search problem in scenarios involving static obstacles and dynamic escape targets, modeling the problem within the framework of Decentralized Partially Observable Markov Decision Process. Based on this model, a Spatio-Temporal Efficient Exploration network and a Global Convolutional Local Ascent mechanism are proposed. Subsequently,we introduce a Multi-UAV Escape Target Search Algorithm Based on MAPPO(ETS-MAPPO) for addressing the escape target search difficulty problem. Simulation results demonstrate that the ETS-MAPPO algorithm outperforms five classical MARL algorithms in terms of the number of target searches,the area coverage rate, and other metrics.

The land use land cover (LULC) map is extensively employed for different purposes. The LULC maps provided by the Portuguese National Geographic Information System (SNIG) are not ideal for continuous assessment and regular scrutiny. Machine learning (ML) algorithms applied in remote sensing (RS) data has been proven effective in image classification, object detection, and semantic segmentation. Previous studies have shown that Random Forest (RF) and Support Vector Machine (SVM) consistently achieved high accuracy for land classification. In this study, the application of RF and SVM classifiers, and object-based (OBIA) and pixel-based (PBIA) approaches with Sentinel-2A imagery was evaluated using Google Earth Engine (GEE) platform for land cover classification of a burned area in the Serra da Estrela Natural Park (PNSE), Portugal. This aimed to detect the land cover change and closely observe the burned area and vegetation recovery after the 2022 wildfire. The combination of RF and OBIA achieved the highest accuracy in all evaluation metrics. At the same time, a comparison with the Normalized Difference Vegetation Index (NDVI) map and Conjunctural Land Occupation Map (COSc) of 2023 year indicated that the SVM and PBIA map resembled the maps better.

Rift Valley Fever (RVF), is a viral zoonotic disease predominant in East Africa and transmitted by Aedes mosquitoes carrying the virus. Using the systematic literature review approach, the present study evaluated machine learning techniques and time series approaches to find literature on the impact of climatic changes on RVF outbreaks published between 1930 and 2024. The literature search involved databases including PubMed, PLOS ONE, JSTOR, Web of Science, Google Scholar, and SCOPUS (Kenmoe et al., 2023).The results show that most of the articles were published between 2018 and 2022, and most of the articles were from United States, France, and Kenya. We conducted a detailed review of the articles using the PRISMA 2020 flow chart, screening and qualifying 10,015 articles. Some articles revealed significant gaps in both internal and external validation. Therefore, future research should focus on developing multi-disciplinary models that incorporate climatic condition, geographical, biological, and social factors.

Falls are a major public health problem among older adults, therefore predicting the risk of having falls in the near future is of great importance for health and social systems worldwide. Nowadays, prospective fall risk assessment relies on clinical and functional mobility assessment tools, such as the Timed Up and Go (TUG) test. Recently, wearable inertial measurement unit (IMU) sensors measurements have been proposed for fall detection. We hypothesize that the IMU readings captured during TUG test realizations can provide prospective fall risk prediction. In this study, analysis by deep learning convolutional neural networks (CNN) and recurrent neural networks (RNN) were used for fall prediction based on features extracted from IMU data acquired during TUG test realizations. Data is obtained from a cohort of 106 older adults wear-ing a wireless inertial sensor with a sampling frequency of 100 Hz while performing the TUG test. Prospective fall incidence was obtained in a six-month follow-up period and used as the ground truth for the supervised training and validations of the deep learning and competing machine learning approaches. Further data collection could lead to potential fall risk biomarker. A repeated hold out cross-validation process using 75 subjects for training and 31 subjects for testing was carried out. Best results were achieved by a bidirectional long short-term memory (BLSTM), obtaining an accuracy of 0.83 and AUC of 0.73 with good sensitivity and specificity values.

(1) Background: Efficient detection and rectification of defects or post-processing manufacturing conditions such as sharp edges and burrs on metal components are crucial for quality control in precision manufacturing industries. (2) Methods: This paper describes a lab-scale integrated system for real-time and automated metal edge image detection using a customized YOLOv5 ma-chine vision algorithm. The YOLOv5 algorithm and model were developed and embedded in the NVIDIA Jetson Nano microprocessor. The YOLOv5 model can detect three different image types for classification i.e., normal, sharp, and burrs edge on aluminum workpiece blocks. An integrated system connects the NVIDIA Jetson Nano with an embedded YOLOv5 model to a Mitsubishi Electric Melfa RV-2F-1D1-S15 manipulator to perform selective automated chamfer-ing and grinding when certain condition defects are detected. (3) Results: The model demonstrates durable performance, achieving a mean average precision of 0.886 across defect classes with minimal misclassifications. The Mitsubishi Electric Melfa RV-2F-1D1-S15 manipulator received input from the machine vision system and performed an automated chamfering and grinding process accordingly; (4) Conclusions: By integrating camera, embedded deep learning in the microprocessor and manipulator, automated chamfering and grinding process in metal component shapes can be efficiently rectified. This tailored solution promises to improve productivity and consistency in metal manufacturing and remanufacturing.

Lung and colon cancers are among the leading causes of cancer-related mortality worldwide. Early and accurate detection of these cancers is crucial for effective treatment and improved patient outcomes. False or incorrect detection is more harmful. Accurately detecting cancer in a patient's tissue is crucial to their effective treatment. While analyzing tissue samples is complicated and time-consuming, deep learning techniques have made it possible to complete this process more efficiently and accurately. As a result, researchers can study more patients in a shorter amount of time and at a lower cost. Much research has been conducted to investigate deep learning models that require great computational ability and resources. However, None of these have had a 100% accurate detection rate for these life-threatening malignancies. Misclassified or falsely detecting cancer can have more harmful consequences. This research proposes a new lightweight, parameter-efficient, and mobile-embedded deep learning model based on a 1D convolutional neural network with Squeeze-and-Excitation layers for efficient lung and colon cancer detection. This proposed model diagnoses and classifies lung squamous cell carcinomas and adenocarcinoma of the lung and colon from digital pathology images. Extensive experiment demonstrates that our proposed model achieves 100% accuracy for detecting lung, colon, and lung and colon cancers from the histopathological (LC25000) lung and colon datasets, which is considered the best accuracy for around 0.35 million trainable parameters and around 6.4 million flops. Compared with the existing results, our proposed architecture shows state-of-the-art performance in lung, colon, and lung and colon cancer detection.

The COVID-19 pandemic has highlighted the need for a robust medical infrastructure and crisis management strategy as part of smart-city applications, with technology playing a crucial role. The Internet of Things (IoT) has emerged as a promising solution, leveraging sensor arrays, wireless communication networks, and artificial intelligence (AI)-driven decision-making. Advancements in Edge Computing (EC), Deep Learning (DL), and Deep Transfer Learning (DTL) have made IoT more effective in healthcare and pandemic-resilient infrastructure. When combined with medically oriented IoT setups, DL-architectures are suitable for integrating into pandemic-compliant medical infrastructure. The development of intelligent pandemic-compliant infrastructure requires combining IoT, edge and cloud computing, image processing, and AI tools to monitor adherence to social distancing norms, mask-wearing protocols, and contact tracing. The proliferation of 5G wireless communication has enabled ultra-wide broadband wireless communication, with high reliability and low latency, thereby enabling seamless medical support as part of smart-city applications. Such set-ups are designed to be ever-ready to deal with virus-triggered pandemic-like medical emergencies. This study presents the design of a pandemic-compliant mechanism leveraging IoT optimized for healthcare applications, edge- and cloud-computing frameworks, and a suite of DL-tools. The framework uses a composite attention-driven framework incorporating various DL-pre-trained models (DPTM) for protocol adherence and contact tracing. When connected to public networks, it can detect specific cyber-attacks. The results confirm the effectiveness of the proposed methodologies.

In the dynamic realm of agribusiness construction, firms are striving to bolster their operational efficiency and profitability amidst the global shift from traditional farming to commercial agriculture. This transition has intensified the demand for sophisticated infrastructure development, presenting new challenges for commercial managers. A critical hurdle is the accurate estimation of profitability for prospective contracts, a task often reliant on intuition rather than data-driven methods. To address this, we propose the development of a mathematical model utilising machine learning techniques to predict contract profitability and identify influential factors. This model aims to aid in bid decision-making, financial forecasting, and enhancing competitiveness in the marketplace. Furthermore, it would provide valuable insights into how altering specific contract attributes might affect predicted profitability. The implementation of such a system necessitates close collaboration between IT professionals and construction executives. This study delineates the development of this predictive model, outlining the data analysis process and the application of machine learning algorithms to tackle this complex commercial challenge. The ultimate goal is to bridge the gap between intuitive estimation and data-driven prediction, thereby enhancing the financial performance and strategic decision-making capabilities of firms in the agribusiness construction sector.

This research centres on advancing the utilisation of a novel Deep Ensemble model called RicEns-Net for predicting crop yields through sophisticated multi-modal data engineering. Initially, data for the study were acquired from Ernst \& Young (EY), a prominent global technology and management consulting firm, as part of their annual data science challenge (EY Open Science Challenge 2023). The predictive models leverage multi-modal data by integrating radar and optical-type remote sensing data alongside meteorological data to create supplementary features essential for predicting crop yield. The primary data features are derived from spectral indices extracted from Sentinel 2 satellites, while more advanced features are generated through combinations and interactions of these fundamental features. Furthermore, data from Sentinel 1 and 3, along with meteorological measurements such as surface temperature and rainfall, account for the intricate natural factors influencing crop yield, creating a comprehensive set of data features based on existing research. To tackle the complexity introduced by numerous data features, extensive data engineering was conducted to identify the most predictive features among 100+ and mitigate the "curse of dimensionality," resulting in 15 features from 5 different modalities. The proposed RicEns-Net comprises multiple machine learning algorithms in a deep ensemble approach to leverage the best features of each of these techniques, achieving optimal performance. The proposed model achieves an MAE value of 341 kg/Ha for prediction, surpassing the performance of the EY Open Science Challenge 2023 winner models and state-of-the-art deep/machine learning approaches.

Background/Objectives: The COVID-19 pandemic reduced in-person pediatric visits in the United States by over 50%, while telehealth visits increased significantly. The national use of telehealth for children and the factors influencing its use have been rarely studied. This study aimed to investigate the prevalence of telehealth use during the COVID-19 pandemic and explore the potential factors linked to its use at the state level. Methods: A cross-sectional study of the National Survey of Children’s Health (2021-22) sponsored by the federal Maternal and Child Health Bureau was performed. We used the least absolute shrinkage and selection operator (LASSO) regression to predict telehealth use during the pandemic. A bar map showing the significant factors from the multivariable regression was created. Results: Of the 101,136 children, 15.25% reported using telehealth visits due to COVID-19, and 3.67% reported using telehealth visits due to other health conditions. The Northeast states showed the highest telehealth use due to COVID-19. In the Midwest and Southern states, children had a lower prevalence of telehealth usage unrelated to COVID-19. The LASSO regressions demonstrated that telehealth usage (due to and not due to) was associated with age, insurance type, household income, usual source of pediatric preventive care, perceived child health, blood disorders, allergy, brain injury, seizure, ADHD, anxiety, depression, and special needs. Conclusions: This study demonstrated significant variability in the use of telehealth among states during the COVID-19 pandemic. Understanding who uses telehealth and why, as well as identifying access barriers, helps maximize telehealth potential and improve healthcare outcomes for all.

While the surface of the Earth plays a key role in weather forecasting through its interaction with the atmosphere, in ensemble numerical weather predictions the uncertainty on the surface is only represented with perturbations in the parameterisations representing the surface processes. Data representing the surface, such as the land cover, are not perturbed. As fully data-driven forecasts without parameterisations are growing in importance, sampling the uncertainty on the land cover data brings a new way of making ensemble forecasts. Our work describes a method of generating ensemble land cover maps for numerical weather prediction. The target land cover map has the ECOCLIMAP-SG labels, used in the SURFEX surface model, and therefore is expected to have all relevant labels for surface-atmosphere interactions. The method translates the ESA WorldCover map to ECOCLIMAP-SG labels and resolution using auto-encoders. The land cover ensemble members are obtained by sampling the land cover probabilities in the output of the neural network. This paper builds upon the work done in a companion paper describing the high-resolution version of ECOCLIMAP-SG, called ECOCLIMAP-SG+, used for the training and evaluation of the neural network. The output map presented here, called ECOCLIMAP-SG-ML, improves upon the ECOCLIMAP-SG map in terms of resolution (from 300~m to 60~m), overall accuracy (from 0.41 to 0.63) and the ability to produce ensemble members.

Accurate forecasting and efficient management of commodities are crucial for stakeholders in the industry, given the volatile nature of these markets. This study addresses these needs by leveraging advanced forecasting techniques and machine learning models to predict prices and enhance supply chain efficiency. TThe focus is on utilizing ARIMA, SARIMAX, and LSTM models to analyze historical trading data for commodities such as cocoa, coffee, and sugar sourced from Kaggle’s comprehensive dataset. The research applies ARIMA and SARIMAX models to forecast price trends, overcoming initial challenges related to data index frequency and seasonality. LSTM models are employed for more nuanced demand forecasting, particularly for random-length lumber, demonstrating the model's capability to predict future market trends accurately. The study highlights significant improvements in prediction accuracy and supply chain management through meticulous feature engineering and model optimization, offering valuable insights for strategic decision-making in the supply chain sector.

Objective: Despite long simulation times, recently developed finite element analysis (FEA) models of knee joints have demonstrated their suitability for predicting individual risk of onset and progression of knee osteoarthritis. Therefore, the objective of this study was to assess the feasibility of machine learning (ML) to replicate outcomes obtained from FEA when simulating mechanical responses and predicting cartilage degeneration within knee joint. Design: Two ML models based on the Gaussian Process Regression (GPR) algorithms were developed. The first model (GPR1) utilized age, weight, and anatomical joint dimensions as predictor variables to predict tissue mechanical responses and cartilage degeneration based on FEA data. The second model (GPR2) utilized age, weight, height, and gender to predict anatomical joint dimensions, which were then used as inputs in the GPR1 model. Finally, the GPR1 and combined GPR1+GPR2 models were used to investigate the importance of clinical imaging when making personalized predictions for knees from healthy subjects with no history of knee injuries. Results: In the GPR1 model, R2 of 0.9 was exceeded for most of the predicted mechanical parameters. The GPR2 model was able to predict knee shape with R2 of 0.67 - 0.9. Both GPR1 and combined GPR1+GPR2 models offered equally good performance (AUC = 0.73 - 0.74) at classifying patients at high risk for the onset and development of knee osteoarthritis. Conclusions: In the future, real-time and easy-to-use GPR models may provide a rapid technology to evaluate mechanical responses within knee for researchers or clinicians who have no former knowledge from FEA.

Breast cancer is the most prevalent type of disease among women. It has become one of the foremost causes of death among women globally. Early detection plays a significant role in administering personalized treatment and improving patient outcomes. Mammography procedures are often used to detect early-stage cancer cells. This traditional method of mammography while valuable has limitations of potential for false positives and negatives, patient discomfort, and radiation exposure. Therefore, there is a probe for more accurate techniques required in detecting breast cancer, leading to exploring the potential of machine learning in the classification of diagnostic images due to its efficiency and accuracy. This study conducted a comparative analysis of pre-trained CNNs (ResNet50 and VGG16) and Vision Transformers (ViT-Base and SWIN Transformer) with the inclusion of ViT-Base trained from scratch model architectures to effectively classify breast cancer mammographic images into benign and malignant cases. The Swin transformer exhibits superior performance with 99.8% accuracy and a precision of 99.8%. These findings demonstrate the efficiency of deep learning to accurately classify breast cancer mammographic images for the diagnosis of breast cancer, leading to improvement in patient outcomes.

This study aimed to identify whether the combined use of functional electrical stimulation (FES) reduces the motor torque of a gait exercise rehabilitation robot in spinal cord injury (SCI) and to verify the effectiveness of the developed automatic assist level adjustment in people with paraplegia. Acute and chronic SCI patients (1 case each) performed 10 min of gait exercises with and without FES using a rehabilitation robot. Reinforcement learning was used to adjust the assist level automatically. The maximum torque values and assist levels for each of the ten walking cycles when walking became steady were averaged and compared with and without FES. The motor's output torque and the assist level were measured as outcomes. The assist level adjustment allowed both the motor torque and assist level to decrease gradually to a steady state. The motor torque and the assist levels were significantly lower with the FES than without the FES under steady conditions in both cases. No adverse events were reported. The combined use of FES attenuated the motor torque of a gait exercise rehabilitation robot for SCI. Automatic assistive level adjustment is also useful for spinal cord injuries.

Maritime transport has adapted to recent political and economic shifts by addressing stringent pollution reduction requirements, redrawing transport routes for safety, reducing onboard technical incidents, managing data security risks and transitioning to autonomous vessels. The integration of AI-based technologies for fault diagnosis and decision-making, coupled with proper personnel training, has become essential for optimal ship operation, with AI-assisted predictive maintenance emerging as the key strategy to ensure onboard systems function within desired parameters and enhance equipment availability. This study proposes a machine-learning algorithm for evaluating onboard systems performance by analyzing sensor data and mapping it to fault patterns to estimate functional states, validated through tests on a seawater cooling system from an oil tanker, demonstrating operational efficiency and reliability. The immediate advantage is reduced time for fault diagnosis by rapidly assessing the system using operational data, the algorithm enabling real-time monitoring and fault data management to minimize efforts as part of predictive maintenance assisted by machine learning tools.

Predicting credit default risk is important to financial institutions, as accurately predicting the likelihood of a borrower defaulting on their loans will help to reduce financial losses, thereby maintaining profitability and stability. Although machine learning models have been used in assessing large applications with complex attributes for these predictions, there is still a need to identify the most effective techniques and to also address the issue of data imbalance. In this research, we conducted a comparative analysis of random forest, decision tree, SVM (Support Vector Machine), XGBoost (eXtreme Gradient Boosting), ADABoost (ADAptive Boosting) and multilayered perceptron with three-hidden layers, to predict credit default using loan data from LendingClub. Additionally, we also combined the model predictions using voting and stacking ensemble methods to enhance the models' performance. Furthermore, various sampling techniques was explored to handle the issue of class imbalance observed in the dataset, with the result showing that the balanced data performs better than the imbalanced data. Our proposed model achieved an accuracy of 93.7%, a precision of 95.6% and a recall of 95.5%, which shows the potential of ensemble methods in improving credit default predictions and can provide lending platforms with the tool to reduce default rates and financial losses. In conclusion, the findings from this study have broader implications for financial institutions, offering a robust approach to risk assessment beyond the Lend-ingClub dataset.

The fundamental period is one of the most important parameters for the design of new structures , as well as the estimation of the capacity of existing ones. However, its calculation requires the solution of an eigenvalue problem. Especially for new buildings, this can involve material and geometric parameters that are not known beforehand. Thus, to estimate it, various design codes and researchers have adopted a several approximate analytical equations based on a number of key structural parameters. To this end, the present study introduces a novel methodology for the derivation of analytical equations for the fundamental period of Reinforced Concrete (RC) structures. The methodology is based on Machine Learning (ML) explainability techniques and specifically the so-called SHapley Additive exPlanations (SHAP) values. The novel methodology allows these equations to be constructed sequentially and in an informed manner, controlling the balance between accuracy and complexity. An extended dataset comprised of 4026 data points is employed. The results showed excellent accuracy, with a coefficient of determination R2≈0.95. This demonstrates the potential applicability of the proposed methodology in a wide array of similar engineering challenges.

This study explores the industrial property market in Hong Kong, a sector characterized by its unique features and atypical market behaviour. Particularly, this research employs machine learning techniques to predict property prices based on a set of features, including location, square footage, floor level, accessibility to mass transit railway station and the like. To ensure transparency and understanding, the Shapley value is employed to quantify the relative importance of each feature in predicting property prices. Our analysis reveals the existence of non–linear relationships among these features, as demonstrated by the wide distribution of SHAP values for most features, which are illustrated in a beeswarm plot that span both sides of the baseline. This finding indicates a complex interaction among such features as square footage, age, floor level, carpark, proximity to mass transit railway stations, location, and property prices. The results contribute valuable insights into the relationships between industrial property characteristics and their corresponding values, thereby equipping stakeholders with enhanced understanding of the market to support informed decision–making.

The spread of false information can significantly harm public opinion, underscoring the importance of accurately identifying untrustworthy news. This paper presents an innovative machine learning (ML) tool, TMining, designed to evaluate news credibility and facilitate various text-mining tasks. By examining a range of ML methodologies alongside preprocessing techniques, we aim to boost the system's effectiveness. Our research meticulously assesses different datasets, highlights the impact of applying stemming techniques, and employs Local Interpretable Model-Agnostic Explanations (LIMEs) to shed light on the rationale behind model predictions. The outcomes reveal a notable enhancement in both the precision and clarity of the news verification process. The ultimate version of the model has been made available as an Application Program Interface (API), and its source code has been shared openly, encouraging further exploration and collaboration within the scientific community. This initiative advances our ability to discern manipulative context from fictitious content and promotes transparency and understanding in the domain of ML applications.

Background: Diabetic foot sepsis (DFS) accounts for approximately 60% of hospital admissions in patients with diabetes mellitus (DM). Individuals with DM are at risk of severe COVID-19. This study investigated investigated factors associated with major amputation and mortality in patients admitted with DFS during the COVID-19 pandemic. Methods: Demographic information, COVID-19 and HIV status, clinical findings, laboratory results, treatment and outcome from records of patients with diabetic foot sepsis. The chi-square test, Fisher’s test, Student t-test or Mann-Whitney compare findings when appropriate. Multivariate logistic regression to determine factors that were associated with major amputation and mortality. Results of logistic regression were compared with outputs from analysis using machine learning algorithms. Kaplan-Meier survival curve to compare survival between COVID-19 positive and negative patients. Statistical was a p-value below 0.05. Supervised machine learning algorithms were used to compare their ability to predict major and deaths. Results: Overall, 114 records were found and 57.9% (66/114) were of male patients. The mean age of the patients was 55.7 (14) years and 47.4% (54/114) and 36% (41/114) tested positive for COVID-19 and HIV, respectively. The median c-reactive protein was 168mg/dl, urea 7.8mmol/l and creatinine 92µmol/l. The mean potassium level was 4.8±0.9mmol, and glycosylated haemoglobin 11.2±3%. The main outcomes included major amputation in 69.3% (79/114) and mortality of 37.7% (43/114) died. AI. The level of potassium, urea, creatinine and HbA1c were significantly higher in the deceased. Conclusions: The COVID-19 pandemic led to an increase in the rate of major amputation and mortality in patients with DFS. The in-hospital mortality was higher in patients above 60 years of age who tested positive for COVID-19. The Random Forest algorithm of ML can be highly effective predicting major amputation and death in patients with DFS.

This study presents the development and evaluation of a Multi-Task Long Short-Term Memory (LSTM) model with an Attention Mechanism designed to predict students' academic performance. The model concurrently addresses two tasks: predicting overall performance (total score) as a regression task and categorizing performance levels (remarks) as a classification task. By processing both tasks simultaneously, the model optimizes computational efficiency and resource use. The dataset includes detailed student performance records across various metrics such as Continuous Assessment, Practical Skills, Demeanor, Presentation Quality, Attendance, and Participation. The model's performance was evaluated using comprehensive metrics. For the regression task, it achieved a Mean Absolute Error (MAE) of 0.0249, Mean Squared Error (MSE) of 0.0012, and Root Mean Squared Error (RMSE) of 0.0346. For the classification task, it attained perfect scores with an accuracy, precision, recall, and F1 score of 1.0. These results highlight the model's high accuracy and robustness in predicting both continuous and categorical outcomes. The Attention Mechanism enhances the model's capabilities by identifying and focusing on the most relevant features. This study demonstrates the effectiveness of the Multi-Task LSTM with Attention Mechanism in educational data analysis, offering a reliable tool for predicting student performance and potential broader applications in similar multi-task learning contexts. Future work will explore further enhancements and wider applications to improve predictive accuracy and efficiency.

The emergence of few-shot learning as a potential approach to address the problem of data scarcity by learning underlying pattern from a few training sample had so far given a mix-result especially on the suitability of model-agnostic meta learning, transfer learning, and optimization strategy to rapidly learn valid information from few sample. In this research, we did an in-depth evaluation of meta- learning to determine their plausibility and suitability for previously unknown cyberattack detection by first retrieving the original research artifacts of current state of the art meta learning to repeat the experiment with original dataset before replicating the experiment with two different malware dataset which had not been previously done with meta-transfer learning. On each of the experiments, meta-transfer learning gave good results on digital character recognition dataset but abysmal result on Malimg and Malevis malware images datasets thereby indicating its unreliability for detecting cyberattacks and the need for an improvement to the state-of-the-art meta transfer learning towards a better attack detection. Transfer learning performance is independent on imbalance and hence does not influence its performance since both malware dataset used for this experiment result in high validation loss and balancing the dataset doesn't result in reduced validation loss,the successful learning transfer seen on digital character recognition dataset is not unconnected to the fact that several languages have similar characters and digits thereby enhancing the successful learning transfer unlike malware datasets, and more importantly the finding that current meta-learning transfer approach doesn't generalize well on malware dataset and hence not suitable for detecting previously unseen out-of-distribution attack.

Streamflow forecasting is of great importance in water resources management and flood warnings. Machine learning techniques can be utilized to assist with river flow forecasting. By analyzing historical time series data on river flows, weather patterns, and other relevant factors, machine-learning models can learn patterns and relationships to present predictions about future river flows. In this study, an Autoregressive Integrated Moving Average (ARIMA) model is constructed to predict the monthly flows of the Athabasca River at three monitoring stations: Hinton, Athabasca, and Fort MacMurray, in Alberta, Canada. The three monitoring stations upstream, midstream, and downstream were selected to represent the different climatological regimes of the Athabasca River. Time series data were used for the model training to identify patterns and correlations using moving averages, exponential smoothing, and Holt-Winters' method. The model's forecasting was compared against the observed data. The results show that the determination coefficients were 0.99 at all three stations, indicating strong correlations. The root mean square errors (RMSEs) were 26.19 at Hinton, 61.1 at Athabasca, and 15.703 at Fort MacMurray, respectively, and the mean absolute percentage errors (MAPEs) were 0.34%, 0.44%, and 0.14%, respectively. Therefore, the ARIMA model captured the seasonality patterns and trends in the stream flows at all three stations and demonstrated a robust performance for hydrological forecasting. This provides insights and predictions for water resources management and flood warnings.

Background: Ensemble tree-based models such as Xgboost are highly prognostic in cardiovascular medicine, as measured by the Clinical Effectiveness Metric (CEM). However, their ability to handle correlated data, such as hospital-level effects, are limited. Objectives: The aim of this work is to develop a binary outcome mixed effects Xgboost (BME) model that integrates random effects at the hospital level. To ascertain how well the model handles correlated data in cardiovascular outcomes, we aims to assess its performance and compare it to fixed effects Xgboost and traditional logistic regression models. Methods: A total of 227,087 patients over 17 years of age, undergoing cardiac surgery from 42 UK hospitals between 1 Jan 2012 and 31 Mar 2019 were included. The dataset was split into two cohorts: Training/Validation (n = 157196; 2012-2016) and Holdout (n = 69891; 2017-2019). The outcome variable was 30 days mortality with hospitals considered as clustering variable. The logistic regression, mixed effects logistic regression, Xgboost and binary outcome mixed effects Xgboost (BME) were fitted to both standardized and unstandardized datasets across a range of sample sizes and the estimated prediction power metrics were compared to identify the best approach. Results: The exploratory study found high variability in hospital-related mortality across datasets, which supported the adoption of mixed effects models. Unstandardized Xgboost BME demonstrated marked improvements in predictor power over the Xgboost model at small sample size ranges, but performance differences decreased as dataset sizes increased. Generalized linear models (glm) and generalized linear mixed-effects models (glmer) models followed similar results, with Xgboost models excelling also at greater sample sizes. Conclusions: These findings suggest that integrating mixed effects into machine learning models can enhance their applicability in clinical settings where sample size is small such as rare conditions.

Background/Objectives: Traditional neurosurgical training involves a steep learning curve. The introduction of advanced simulation technologies, such as virtual reality (VR), provides an alternative method for skill acquisition, allowing for repeated practice and objective assessment. This study focuses on evaluating the learning curve associated with lumbar pedicle Kirschner wire insertion using a virtual fluoroscopic simulator among neurosurgical residents. Methods: Eighteen neurosurgery residents and two interns participated in this study. Participants’ initial skill levels were assessed through a questionnaire. Performance metrics, including insertion accuracy, time, and the number of virtual X-ray scans, were recorded across multiple attempts until a “practical learning plateau” was reached. Statistical analyses were conducted to model learning curves, assess the correlation between pre-training experience and performance. Results: The average number of attempts required to stabilize performance was 5.61. Learning rates varied, with some residents showing rapid improvement while others progressed more slowly. Higher accuracy in wire placement correlated with reduced usage on X-rays over time. However, experienced residents did not necessarily learn faster, indicating potential challenges in adapting to new simulation-based methods. Post-training feedback highlighted the simulator’s utility in increasing confidence and skill levels, although some limitations in anatomical accuracy were noted. Conclusions: This study demonstrates the potential of virtual simulation to enhance neurosurgical training by providing a controlled environment for repeated practice and objective feedback. Simulation-based training can effectively complement traditional methods, though individualized approaches may be necessary to accommodate varying learning rates among residents. Further refinement of simulation tools and their integration into standard training curricula are recommended.

This article compares the predictive capabilities of six models namely the Linear Discriminant Analysis (LDA), Logistic Regression (LR), Support Vector Machine (SVM), XGBoost, Random Forest (RF) and Deep Neural Network (DNN) to predict the default behaviour of credit card holders in Taiwan using data from the UCI machine learning database. Python programming language was used for data analysis. Statistical methods were compared with machine learning algorithms using the confusion matrix measured in metric terms of prediction accuracy, sensitivity, specificity, precision, G-mean, F1 score, ROC and AUC. The dataset contains 30,000 credit card user’s information with 6636 default observations and 23,364 non-default cases. The study results found that modern machine learning methods outperformed traditional statistical methods in terms of predictive performance measured by F1 score, G-mean and AUC. Traditional methods like logistic regression were marginally better than linear discriminant analysis and support vector machines in terms of predictive performance measured by area under the receiver operating characteristic curve. In the modern machine learning methods, deep neural networks were better in most of the predictive performance metrics than XGBoost and Random Forest methods.

Deep Reinforcement Learning has achieved state-of-the-art results in the field of autonomous driving, especially in video game simulations. Trained agents were able to beat the best players achieving supremacy in the field. However, these agents often take in their neural network input the geometry of the track during the learning process. Our approach is to create a learning procedure for autonomous driving agents that is not environment-dependent by design to extend Deep Reinforcement Learning to racing games without racetrack information. Our work builds upon the notorious game Trackmania Nations United Forever, an open access 3D realistic Formula 1 simulation game that offers a high degree of customization. We use the Proximal Policy Optimization algorithm, using hybrid raw pixel data and physics sensors vectors as agent observations. The training track is a custom-made circuit carefully crafted from all available blocks in the game. The training track is considered like a standard dataset in machine learning and thus split in several segments in which the agent is randomly put into. We also created two regularization reward functions for the car’s speed and position to increase robustness of the agent in new scenarios.

In this digital era, online learning and teaching is an important platform for college and university learners which required many technical facilities and in depth knowledge to have learning process effectively. Students and teachers in Higher Education Institutions (HEIs) were forced to grapple with the inevitability of shifting to online classes due to the advantages of effective distance learning or for large learning group. In the case of Oman, there have been some efforts to integrate alternative learning mechanisms such as online and mobile education in the past but not followed the scale in the post pandemic world. This study seeks to utilize the principles of Total Quality Management (TQM) through the implementation of Quality Function Deployment (QFD) as a way to better understand the experiences of Omani teachers and students from HEIs. A voice of customer from teachers and students have been collected and recorded. In this regard, 165 staff and 355 students from HEIs of Engineering participated in the survey. After the survey, a house of quality analysis was done to analyze the requirements based on how strongly they are related to the standard learning outcomes expected from HEIs in Oman. The results show that more attention is required in online education in the Sultanate of Oman before it is implemented as a major part in learning and teaching.

Information dissemination and preservation are crucial for societal progress, especially in the technological age. While technology fosters knowledge sharing, it also risks spreading misinformation. Audio deep fakes, convincingly fabricated audio using Artificial intelligence (AI), exacerbate this issue. We present Sonic Sleuth, an AI model for detecting audio deepfakes. Our approach leverages Deep Learning (DL) for a robust detection model. Meticulous data preprocessing and rigorous experimentation with various models led to the implementation of the most effective solution with a custom CNN model. Our comprehensive testing resulted in a highly accurate model (98.27% accuracy, 0.016 EER) trained on a substantial dataset of real and synthetic audio. In addition to an 84.92% accuracy and 0.085 EER on an external dataset, these results demonstrate Sonic Sleuth’s potential as a powerful tool against audio misinformation.

As the adoption of deep learning models continues to surge across various applications, the need for efficient deployment architectures becomes increasingly critical. This paper presents a novel approach to enhance the deployment of deep learning models by leveraging serverless architecture. Serverless computing has been popular for its auto-scaling, cost-effectiveness, and simplified management characteristics. However, the intense resource demands of deep learning models pose challenges in maintaining low response times and effective load balancing within serverless environments. The proposed architecture addresses these challenges by integrating principles from both deep learning model optimization and serverless computing. Through systematic experimentation and analysis, we demonstrate that by appropriately designing and tuning the deployment architecture, significant improvements in response time, performance, resource utilization, and load distribution can be achieved.

The last wave of Artificial Intelligence is dedicating particular emphasis to the line of explainability research. The potential of Explainable Artificial Intelligence (XAI) in producing trustworthy computer-aided-diagnosis systems and its usage for knowledge discovery are collecting interest in the Medical Imaging (MI) community to support the diagnostic process and the discovery of image biomarkers. Most of the existing XAI applications in MI were focused on interpreting the predictions returned by deep neural networks, typically including attribution techniques with saliency map approaches and other feature visualization methods. However, these are often criticized for providing incorrect and incomplete representations of the black-box models’ behaviour. This raises the attention in proposing models intentionally designed to be self-explanatory. In particular, part-prototype (PP) models are interpretable-by-design computer vision (CV) models that base their decision process in learning and identifying representative prototypical parts from the input images, and they are collecting increasing interest and results in MI applications. This narrative review provides a summary of existing PP networks, their application in MI analysis and current challenges.

This paper investigates the applicability of deep learning models for predicting the severity of forest wildfires, utilizing an innovative benchmark dataset called EO4WildFires. EO4WildFires integrates multispectral imagery from Sentinel-2, SAR data from Sentinel-1 and meteorological data from NASA Power annotated with EFFIS data for forest fire detection and size estimation. Resulting in a coverage of 45 countries (Figure \ref{Figure1-events}) with a total of 31730 wildfire events from 2018 to 2022. All these various sources of data are archived into data cubes, with the intention of assessing wildfire severity by considering both current and historical forest conditions, utilizing a broad range of data including temperature, precipitation, and soil moisture. The experimental setup has been arranged to test different deep learning architectures’ effectiveness in predicting the size and shape of wildfire burned areas. The study incorporates both Image Segmentation networks and Visual Transformers, employing a consistent experimental design across various models to ensure comparability of results. Adjustments were made to the training data, such as the exclusion of empty labels and very small events, to refine the focus on more significant wildfire events and potentially improve prediction accuracy. The models’ performance was evaluated using metrics like F1 Score, IoU Score, and Average Percentage Difference (aPD). These metrics offer a multi-faceted view of model performance, assessing aspects such as precision, sensitivity, and the accuracy of the burned area estimation. Through extensive testing with Image Segmentation networks and Visual Transformers, the research not only aims at enhancing the accuracy of estimating the burned area but also underscores the significance of quality training data and the comparative effectiveness of traditional segmentation methods over transformer-based models for this specific application. The dataset is published with an Open Access license and is hosted at: \url{10.5281/zenodo.7762564}.

Differentiated thyroid cancer (DTC), comprising of papillary and follicular thyroid cancers, is the most prevalent type of thyroid malignancy. Accurate prediction of DTC is crucial for improving patient outcomes. Machine learning (ML) offers a promising approach to analyze risk factors and predict cancer recurrence. In this study, we aimed to develop predictive models to identify patients at an elevated risk of DTC recurrence based on 16 risk factors. We developed six ML models and applied them to a DTC dataset. We evaluated the ML models using Synthetic Minority Over-Sampling Technique (SMOTE) and with hyperparameter tuning. We measured the models’ performance using precision, recall, F1 score, and accuracy. Results showed that Random Forest consistently outperformed the other investigated models (KNN, SVM, Decision Tree, AdaBoost, and XGBoost) across all scenarios, demonstrating high accuracy and balanced precision and recall. The application of SMOTE improved model performance and hyperparameter tuning enhanced overall model effectiveness.

It is critical to keep abreast of emerging dangers in the constantly changing field of cyber security. As 2024 approaches, numerous technological developments and trends are changing how businesses handle cyber security. This article explores the cutting edge of cyber security developments that are expected to rule in 2024. The field of cyber security is undergoing a revolution, from the adoption of Zero Trust Architecture (ZTA) as the new paradigm to the integration of AI and ML for real-time threat detection. Solutions for Extended Detection and Response (XDR) are becoming the mainstay for thorough threat detection and response in a variety of contexts, including networks, endpoints, and cloud environments. The use of biometric authentication techniques is growing in popularity because they provide better access control and identity verification. Concurrently, Deception Technologies are altering defensive tactics by using fictitious assets to trick and identify attackers at an early stage. Quantum-resistant cryptography, which aims to protect sensitive data from future quantum attacks, is on the horizon given the threat posed by quantum computing. Furthermore, the growing number of linked devices emphasizes how critical it is to strengthen IoT security protocols. Moreover, as long as companies are using cloud computing, they must make sure that strong cloud security protocols are in place. Simultaneously, the emergence of containerization technologies demands a targeted strategy for containerized environment security. Organizations must adopt these practices and strengthen their defenses against new threats as the cyber security landscape changes. Businesses may confidently and resiliently traverse the complex cybersecurity landscape of 2024 by remaining aware and implementing proactive security measures.

By 2024, artificial intelligence (AI) will have had a revolutionary impact on cyber security, putting the sector at a critical turning point. This article explores the significant influence AI has had on cyber security procedures, outlining the progress made as well as the difficulties faced. artificial intelligence (AI) has transformed threat detection and response systems, allowing for proactive security tactics against a constantly changing array of cyber threats. By utilizing neural networks and machine learning algorithms, artificial intelligence (AI) systems can predict possible security breaches, detect unusual trends, and automatically reduce risks in real-time. These capabilities are unparalleled. Nevertheless, despite these impressive advancements, worries about data privacy, adversarial attacks, and ethical ramifications remain significant. Moreover, the never-ending game of cat and mouse between cyber attackers and defenders continues, requiring constant innovation and adaptability in cyber security strategies driven by Artificial Intelligence. In addition to providing insights into its implications for protecting digital assets and bolstering resilience against cyber threats in the changing cyber security scenario of 2024, this paper summarizes the current state of the artificial intelligence revolution in cyber security.

Currently, the Bangladeshi financial sector is undergoing significant transformations. With the expansion of the infrastructure and the increasing prevalence of digital banking services, it is essential to take measures to prevent cyberattacks. These threats can be effectively mitigated through the utilization of deep learning and machine learning techniques. Through the analysis of massive amounts of banking data in real-time, these methods make it possible to detect and prevent malicious software, fraud, and security breaches that are proactive. Because the government is actively pushing for financial inclusion and digitalization, strong cybersecurity measures are more important than they have ever been. It is the purpose of this study to investigate the application of deep learning and machine learning to improve the cybersecurity environment of Bangladesh’s banking sector in light of the country’s rapidly expanding digital economy.

Organisations need dynamic capabilities in the ongoing digital transformation to reconfigure knowledge and learning. There is a need to define new concepts and explain mechanisms of relevant factors to build dynamic capabilities. Especially organisations acting in healthcare are caught in a dilemmatic situation. New, digital processes and business models are promising benefits for cost-containment measures and improved patient-centric care, offering digital services. However, investments are needed to benefit. The important question is: how can individual actors in healthcare be motivated to engage in this transformational process to build and reconfigure relevant competences and establish new learning routines? Founded on the essence of existing literature, we assume sustained learning as a relevant dynamic capability to seize and sense competences and reconfigure human capital. This paper answers the call for deeper investigations into the mechanisms in new digitally transformed environments and sectors with less focus on performance and competitive advantages, like public administration or the healthcare sector. Based on previous research, validated in qualitative interviews and quantitative testing, we define the new construct of sustained learning with its subdimensions. By providing measures, we build the ground for further quantitative research.

Due to its flexibility, cloud computing has become essential in modern operational schemes. However, the effective management of cloud resources to ensure cost-effectiveness and maintain high performance presents significant challenges. The pay-as-you-go pricing model, while convenient, can lead to escalated expenses and hinder long-term planning. Consequently, FinOps advocates proactive management strategies, with resource usage prediction emerging as a crucial optimization category. In this research, we introduce the multi-time series forecasting system (MSFS), a novel approach for data-driven resource optimization alongside the hybrid ensemble anomaly detection algorithm (HEADA). Our method prioritizes the concept-centric approach, focusing on factors such as prediction uncertainty, interpretability and domain-specific measures. Furthermore, we introduce the similarity-based time-series grouping (STG) method as a core component of MSFS for optimizing multi-time series forecasting, ensuring its scalability with the rapid growth of the cloud environment. The experiments performed demonstrate that our group-specific forecasting model (GSFM) approach enabled MSFS to achieve a significant cost reduction of up to 44%.

In the recent years, with the advent of Artificial Intelligence (AI) traditional methods have seen a significant transformation in the agriculture sector, especially in soil management. Soil management involves practices that maintains and improves the physical, chemical as well as the biological properties of the soil. Soil health management is essential for both the environmental conservation and sustainable agriculture production, ensuring the soil productivity and functional aspects associated with the ecosystem. Soil health management is one of the most important aspects of agriculture and food production, hence preserving and enhancing the soil health is an essential factor for supporting agriculture. Integration of Artificial Intelligence (AI) technologies with soil health management offers the potential to enhance agricultural sustainability, productivity, adapt to the climatic changes and resource constraints. The study of AI tools that can help improve soil health management by providing more accurate and efficient monitoring, analysis and decision-making capabilities. This paper studies the potential AI technologies including machine learning, robotics, and remote sensing in enhancing soil health, raising crop yields, and lowering environmental concerns by examining previous research and case studies.

The Internet of Things (IoT) has revolutionized various industries, but the increased dependence on all kinds of IoT devices and the sensitive nature of the data accumulated by them pose a formidable threat to privacy and security. While traditional Intrusion Detection Systems (IDSs) have been effective in securing critical infrastructures, the centralized nature of these systems raises serious data privacy concerns, as sensitive information is sent to a central server for analysis. This research paper introduces a Federated Learning (FL) approach designed for detecting intrusions in diverse IoT networks that addresses the issue of data privacy by ensuring that sensitive information is kept in the individual IoT devices during model training. Our framework utilizes the Federated Averaging (FedAvg) algorithm, which aggregates model weights from distributed devices to refine the global model iteratively. The proposed model manages to achieve above 90\% accuracy across various metrics, including precision, recall, and f1-score, while maintaining low computational demands. The results show that the proposed system successfully identifies various types of cyber attacks, including DoS, DDoS, Data Injection, Ransomware, and several others showcasing its robustness. This research makes a great advancement to the intrusion detection systems by providing an efficient and reliable solution that is more scalable and privacy friendly than any of the existing models.

In this study, we address the need for improved landslide forecasting by exploring innovative machine learning approaches to classify shallow landslide events. Focusing on Ometepe Island, Nicaragua, known for its susceptibility to hydrometeorological incidents and seismic activities, we constructed a comprehensive shallow landslide database using coherence layers derived from SAR images. We applied coherence thresholds ranging from 0.9 to 0.99, in conjunction with 7-day aggregated spatio-temporal precipitation data. By employing machine learning models such as Logistic Regression, Random Forest, and Support Vector Classifier, we established a correlation between precipitation patterns and landslide occurrences. The Support Vector Classifier with a Sigmoidal kernel proved to be the most effective, achieving an F1 score of 0.67. Our results demonstrate a linkage between precipitation and landslide events, confirming the critical role of rainfall over 7-day periods in triggering these events. The study also indicates that, while rainfall is a crucial factor, it is not the sole trigger for landslides, as seismic activities also play a significant role. This research contributes valuable insights into landslide prediction and risk assessment, underscoring the potential of machine learning in enhancing environmental hazard analysis.

Traffic Inspection (TI) work in this article is positioned as a specific module of road traffic with its primary function oriented towards monitoring and controlling safe traffic and the execution of significant events within a particular geographic area. Exploratory research on the significance of event execution in simple, complicated and complex traffic flow and process situations is related to the activities of monitoring and controlling functional states and performance of categorical variables, which include objects and locations of road infrastructure, communication infrastructure and networks of traffic inspection resources. It is emphasized that the words “work” and “traffic” have the semantic status as synonyms (in one world language), which is explained in the design of the Agent-based model of the complexity of content and contextual structure of TI work at the singular and plural level with 12 points of interest (POI) in the thematic research. An Event Execution Log (EEL) was created for on-site data collection with eight variables, seven of which are independent (event type, activities, objects, locations, host, duration-period and periodicity of the event) and one dependent (significance of the event) variable. The structured data set includes 10,994 input-output vectors in 970 categories collected in the EEL created by 32 human agents (traffic inspectors) over a 30-day period. An algorithmic presentation of the methodological research procedure for preprocessing and final data processing in the ensemble of machine learning models for classification and selection of TI tasks is given. Data cleaning was performed on the available dataset to increase data consistency for further processing. Vector elimination was carried out based on the Location variable, such that the total number of vectors equals the number of unique categories of this variable, which is 636. The main result of this research is the Classification modeling of the significance of events in TI work based on machine learning techniques and the Stacking ensemble. The created machine learning models for event significance classification modeling have high accuracy values. To evaluate the performance metrics of the Stacking ensemble of models, the confusion matrix, Precision, Recall, ND F1-score are used.

Optimization is critical in various fields like smart vehicles, and transportation. Federated Learning (FL) has emerged as an effective approach in the coordination of autonomous vehicles, but traditional methods such as FedAvg can create performance disparities across clients. This paper addresses this fairness issue through the $q$-Fair Federated Learning ($q$-FFL) framework, adjusting model performance across clients using a tunable fairness parameter. We propose a modified FedAvg algorithm for $q$-FFL that maintains comparable convergence rates, ensuring more balanced client outcomes. Additionally, we explore incentive mechanisms in FL using a Stackelberg game model, incorporating a fairness coefficient to encourage equitable participation. Building on prior works, we redefine client utility functions to address communication and computation costs, ensuring fair resource allocation. The proposed framework achieves both global and local fairness, maintaining a unique Nash equilibrium in the modified game setting.

The integration of Genetic Algorithms (GAs) into Serious Games (SGs) has gained traction as a method to optimize game mechanics and personalize user experiences. While GAs are well-known for their effectiveness in solving complex optimization problems, their application in SGs remains relatively under-explored. This review aims to address this gap by systematically mapping the existing literature on the intersection of these two fields. The primary objective of this scoping review is to identify and synthesize the existing research on the application of GAs in SGs, with a focus on understanding the current state of the field, identifying common applications, challenges, and potential future directions. The review includes studies published in English that focus on the use of GAs within SGs. The search encompassed articles from academic journals and conference proceedings without restrictions on publication dates. Exclusion criteria included studies that did not specifically address the integration of GAs in SGs. A systematic search was conducted in databases such as ACM Digital Library, Web of Science, Scopus and Inspec in July 2024. The search terms used included "genetic algorithm," "evolutionary algorithm," "computational algorithm," and "serious games." Unpublished data and articles were not considered. Data were charted using a PRISMA flowchart developed by the author VV. Key categories for data extraction included study objectives, methods, outcomes, type of study and GA used and the specific applications of GAs in SGs. Data charting was conducted independently by two reviewers, with disagreements resolved through discussion and consensus. The search identified 154 studies, of which 23 met the inclusion criteria. The included studies highlight the diverse applications of GAs in SGs, ranging from optimizing game scenarios to personalizing learning experiences. Challenges identified include the computational complexity of GAs and difficulties in their integration into existing game frameworks. The findings of this review suggest that while the use of GAs in SGs is a promising area of research, it remains in its nascent stages. Future research should focus on addressing the technical challenges of integrating GAs into SGs and exploring their application across a wider range of game genres and educational contexts. Limitations of the review include the potential for publication bias and the exclusion of non-English language studies.

N-shot learning has emerge in recent year as potential learning approach to solve the problem of data scarcity by learning underlying pattern from a few training sample. Despite recent state-of-the-art research on model-agnostic metal learning, transfer learning, and optimization strategy to rapidly learn valid information from few sample, there remains a big challenge on an actual out-of-distribution zero-day without any similarity to previously known malware family or new variant of an existing malware family. This ultimately questions the effectiveness of current state-of-the-art few-shot learning approach. In this research, we did an in-depth investigation into the performance of state-of-the-art Zero-shot, Single-shot, and few-shot learning approaches on zero-day out-of-distribution malware attack detection based on their static properties using Malimg and Malevis malware dataset. We ensure our model was aware of an out-of-distribution class during training while varying the number of samples in the out-of-distribution class accordingly zero-shot(no sample), single-shot (1 sample), few-shot(5 samples) while using confusion matrix to get the actual number of correct prediction on out-of-distribution malware validation samples. we assert that the model should be smart enough to detect and classify previously unseen data into an empty family as an out-of-distribution considering that the model was made to be aware of the existence of such distribution during training. Result shows 0, 0, and 3 correct out-of-distribution predictions on Zero-shot, single-shot, and few-shot experiments respectively, thereby showing limitation of the current state-of-the-art N-shot approaches on out-of-distribution attack.

Stunting is a major public health issue in Aceh, Indonesia, requiring advanced analytical techniques for effective interventions. This study presents a novel hybrid machine learning framework designed to enhance the analysis of stunting through improved classification, predic-tive modeling, and clustering optimization. The framework utilizes Support Vector Machines (SVM) with Radial Basis Function (RBF) and Sigmoid kernels for classification. The RBF kernel achieved an accuracy of 91.3%, significantly outperforming the Sigmoid kernel's 85.6%. Linear Regression was employed for predictive modeling, yielding a Mean Squared Error (MSE) of 0.137, which indicates strong predictive accuracy. In clustering, the optimized K-Medoids method, in-corporating a weight product approach, demonstrated superior efficiency by requiring only 3 iterations for convergence, compared to 7 iterations for the conventional K-Medoids method. Additionally, it achieved a higher Calinski Harabasz Index of 93.7, compared to 85.2 for the conventional method. This comprehensive approach enhances accuracy and efficiency across classification, prediction, and clustering tasks, providing valuable insights for targeted interventions and policy development to address stunting in Aceh.

Because the integrity, availability, and confidentiality of digital assets are so important, cybersecurity is key to both the education sector and the critical infrastructure sector. This article investigates the state of cybersecurity in education, emphasizing strategies for securing the classroom against cyberattacks. We go over important subtopics including risk assessment, putting security measures in place, and incident response preparation. Furthermore, we examine how industry-academia partnerships might improve cybersecurity resilience. We also explore cybersecurity for critical infrastructure, stressing the need for risk assessment, security control implementation, and incident reaction and recovery planning. We introduce subtopics such as public-private partnerships in critical infrastructure protection and continuous monitoring and threat detection systems. By use of an extensive examination, this article seeks to provide perspectives on efficacious cybersecurity tactics for preserving education and vital infrastructure, guaranteeing the sustained steadiness and adaptability of the community against constantly changing cyber threats.

Predictive maintenance (PdM) represents a transformative approach to ensuring the security and reliability of critical infrastructure systems. This proactive maintenance strategy leverages advanced data analytics, machine learning, and sensor technology to anticipate and address potential failures before they occur. By continuously monitoring the health of infrastructure components and analyzing performance data, predictive maintenance enables timely interventions, minimizes unplanned downtimes, and enhances overall system resilience. This paper explores the application of predictive maintenance in safeguarding critical infrastructure, focusing on its impact on security, operational efficiency, and cost-effectiveness. Case studies demonstrate how PdM has been successfully implemented across various sectors, including transportation, energy, and utilities, highlighting its role in mitigating risks and extending the lifespan of essential assets. The findings underscore the importance of integrating predictive maintenance into security strategies to protect vital infrastructure from emerging threats and ensure uninterrupted service delivery.

In the rapidly evolving landscape of cybersecurity, Intrusion Detection Systems (IDS) play a critical role in safeguarding networks and systems from unauthorized access and malicious activities. Traditional IDS approaches, relying heavily on predefined rules and signature-based detection, often struggle to keep pace with the dynamic nature of modern cyber threats. Deep learning, with its ability to automatically learn complex patterns and representations from large datasets, offers a promising solution to this challenge. This paper explores the application of deep learning techniques in enhancing the effectiveness and accuracy of IDS. By leveraging advanced neural network architectures such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), and autoencoders, deep learning-based IDS can detect both known and unknown (zero-day) attacks with higher precision. We discuss the advantages of deep learning in IDS, including its capability to handle large-scale data, adaptability to evolving threats, and reduced dependency on human intervention. Furthermore, this paper reviews recent advancements in deep learning for IDS, highlights the challenges associated with their implementation, and suggests future directions for research to overcome these hurdles. The integration of deep learning into IDS frameworks represents a significant step forward in the development of intelligent and autonomous cybersecurity systems.

Recent advancements in Artificial Intelligence (AI) and generative models have significantly accelerated the evolution of internet services. In data centers, where data is often distributed across multiple machines, leveraging these machines to expedite the training of machine learning applications is common. However, this approach incurs substantial communication overhead. To mitigate this, data compression techniques such as quantization, sparsification, and dimension reduction are gaining attention, especially with iterative training algorithms that frequently exchange gradients. With the advancement of FPGA technology, FPGA-based bump-in-the-wire network interface controllers (smartNICs) have demonstrated notable benefits for data centers, matching ASIC switches in port-to-port latency and capacity. However, FPGA-based switches often have spare hardware resources beyond their switching functions, which are typically wasted. we propose a novel architecture that integrates FPGA-based switches with ensemble learning methods, utilizing these spare resources on FPGA-based network switches. This architecture employs random projection as a compression method to accelerate machine learning tasks while maintaining high accuracy. Our system accelerates the training process for Multilayer Perceptron (MLP) models with a speedup of 2.1-6.7 times across four high-dimensional datasets. In addition, it leverages the Hamming distance metric for k-Nearest Neighbors (kNN) classification, combining random projection with Hamming encoding, to offload large-scale classification tasks from downstream servers. This achieves a worst-case speedup of 2.3-8.7 times and a memory reduction of 3.9-27.7 times compared to CPU-based kNN classifiers using cosine distance. Our architecture addresses the challenges of high-dimensional data processing and exemplifies the potential of AI-driven solutions in accelerating and optimizing internet services. Moreover, the programmability of FPGAs allows the system to adopt various compression methods, extending its applicability. Offloading tasks to FPGAs can significantly reduce query response times, showcasing a crucial step towards achieving greater levels of autonomic computing and the rapid evolution of internet systems.

Breast cancer remains a global health problem requiring effective diagnostic methods for early detection in order to achieve WHO’s ultimate goal of breast self-examination (BSE). A literature review indicates the urgency of improving diagnostic methods and identifies thermography as a promising, cost-effective, and non-invasive and adjunctive and complementary detection method. This research explores the potential of using machine learning (ML) techniques, specifically Bayesian Networks (BN) combined with Convolutional Neural Networks (CNN) to improve breast cancer diagnosis at early stages. Explainable artificial intelligence (XAI) aims to clarify the reasoning behind any output of artificial neural networks based models. The proposed integration adds interpretability of the diagnosis, which is particularly significant for a medical diagnosis. We have constructed two diagnostic expert models. In research model A, combining thermal images after XAI process together with medical records, an accuracy of 84.07% has been achieved, while model B, that includes also CNN prediction, achieved an accuracy of 90.93%. These results demonstrate the potential of XAI to transform breast cancer diagnosis, increasing accuracy and reducing the risk of misdiagnosis.

This paper explores the ever-changing topic of cybersecurity and provides a summary of the most recent developments. In a time of unparalleled digital revolution and changing threat environments, being aware of new trends is essential to managing risks and protecting important assets. To highlight significant advancements in fields like artificial intelligence and machine learning for threat detection, the proliferation of IoT devices and related security challenges, the emergence of quantum-resistant cryptography, the development of zero trust architectures, and the growing significance of cybersecurity awareness and education, the report synthesizes insights from industry experts, academic research, and real-world case studies. The goal of the report's analysis of these patterns is to provide stakeholders with useful information so they may modify their cybersecurity plans and strengthen defenses against new threats. Cybersecurity plays a significant role in the IT industry.

Automated Machine Learning (AutoML) is revolutionizing how businesses utilize data, making advanced analytics accessible to a broader range of organizations. By automating complex tasks like data preprocessing, model selection, and hyperparameter tuning, AutoML reduces the time and resources needed to develop and deploy machine learning models. This accelerates decision-making and enables quicker responses to market changes. AutoML empowers businesses to build accurate predictive models using sophisticated algorithms, optimizing model performance for reliable insights and better outcomes. A key advantage of AutoML is its accessibility; even organizations without a dedicated data science team can leverage machine learning, reducing technical barriers and democratizing innovation. As businesses grow, AutoML scales to handle larger datasets and more complex problems without extensive manual intervention. AutoML enhances efficiency, accuracy, and scalability, becoming a crucial driver of business innovation and success. The systematic review will examine the bibliometric literature on how AutoML can boost business, analyzing 74 academic and scientific documents from the Scopus database.

The aim of this article is to discuss the transformative amalgamation of Internet of Things and Machine Learning technologies within the domain of vegetable hydroponic systems for nutrition management. Hydroponics, being an efficient cultivation method, benefits a great deal from the precision and adaptability that machine learning algorithms can offer, along with the real-time monitoring facilitated by the devices using the Internet of Things. This study summarizes the latest research and underlines how ML and IoT may work together, focusing on nutrient optimization, plant development, and resource efficiency. The use of ML algorithms, the function of IoT devices for real-time monitoring, communication protocols, scalability issues, and implementation are among some of the key subjects of the discussion. The research indicates benefits to crop output, efficiency in using resources, and sustainability based on the case studies and the analysis of results. However, ethical problems and some complications concerning data privacy do call for responsible adoptions. The conclusion of this paper provides directions for further research and calls for more investigation into state-of-the-art machine learning approaches and scalable solutions for the re-silient and sustainable future of hydroponic agriculture. The machine learning and deep learning models introduced in this research were evaluated against contemporary studies, revealing an accuracy enhancement ranging from 1.17% to 5.25%, depending on the dataset and algorithm employed. The present study conducts a comparative analysis involving machine learning algo-rithms, indicating that among all the models, the Decision Tree and Gradient Boosting Classifier achieved an accuracy of 99.42% in the dataset by making stage-wise decisions.

Monitoring seaweed growth rates and biomass is crucial for optimizing harvest strategies in aquaculture systems. While such a task can be performed manually on a small farm, such as the Integrated Multi-Trophic Aquaculture (IMTA) system at Harbor Branch Oceanographic Institute at Florida Atlantic University (HBOI), a commercial farm will have to rely on the automated sensor to perform such a task. This study introduces an advanced LSTM-based approach for forecasting seaweed growth and biomass. Utilizing a combination of real and synthetic data, LSTM models are trained and evaluated for their predictive performance. Synthetic sensor data was generated using mathematical equations that simulate realistic aquaculture conditions, with added noise to reflect sensor variability. Building on the foundation of the Pseudorandom Encoded Light for Evaluating Biomass (PEEB) sensor deployed at the seaweed tank in the HBOI IMTA system, we refine the process of biomass estimation by introducing non-linear regression models for predicting seaweed growth and biomass. The results showed that the LSTM model trained with a loss function under physics constraint, combining MSE and physical laws, outperformed models trained with MSE alone, achieving a significantly lower error in predicting seaweed growth. The variation trend of the predicted biomass from the network matched well with the sensor measurement after moving average preprocessing, demonstrating the robustness of the proposed technique in handling noisy sensor data. This study highlights the potential of integrating machine learning with physical models to optimize seaweed cultivation and support sustainable aquaculture practices.

Diabetic retinopathy, a common complication of diabetes, is further exacerbated by factors such as hypertension and obesity. This study introduces the Diabetic Retinopathy Convolutional Transformer (DRCT) model, which combines convolutional and transformer techniques to enhance the classification of retinal images. The DRCT model achieved an impressive average F1 score of 0.97, reflecting its high accuracy in detecting true positives while minimizing false positives. Throughout 100 training epochs, the model exhibited strong generalization capabilities, achieving superior validation accuracy with minimal overfitting. On a newly evaluated dataset, the model attained precision and recall scores of 96.93% and 98.89%, respectively, indicating a well-balanced handling of false positives and false negatives. The model's ability to classify retinal images into five distinct diabetic retinopathy categories demonstrates its potential to significantly improve automated diagnosis and aid in clinical decision-making.

This study investigates the capabilities of high-resolution Sentinel-2 observations and deep learning (DL) segmentation models, specifically convolutional neural networks (CNNs), for accurate mapping of small and fragmented agricultural burn areas in Punjab, India. Initially, the model was trained using ICNF burn area data from Portugal to capture large fire and burn area delineation, achieving moderate accuracy. Subsequent fine-tuning using annotated data from Punjab improved the model’s ability to detect small burn patches, demonstrating higher accuracy and precision than the baseline Normalised Burn Ratio (NBR) Index method. On-ground validation using buffer zone analysis and crop field images confirms the DL approach effectiveness. Cloud interference and temporal gaps in satellite data posed some challenges. Despite these limitations, the study highlights the methodological advancements and potential of using DL models for small burn area detection in agricultural settings. The model achieved overall testing accuracy of 0.64 in Dice score assessment. This report presents a promising approach for improved mapping of the burn area, which might engender better estimation of emissions and the management practices in regions with little available ground truth data.

The quality of wine depends upon the quality of the grapes, which, in turn, are affected by different viticulture aspects and the climate during the grape-growing season. Obtaining wine professionals’ judgments of the intrinsic qualities of selected wine products is a time-consuming task. It is also expensive. Instead of waiting for the wine to be produced, it is better to have an idea of the quality before harvesting, so that wine growers and wine manufacturers can use high-quality grapes. The main aim of the present study was to investigate the use of machine learning aspects in predicting wine quality and to develop a pipeline which represents the major steps from vineyards to wine quality indices. This pipeline outputs the predicted yield, values for basic parameters of grape juice composition, values for basic parameters of the wine composition, and quality. We also found that the yield could be predicted because of input data related to the characteristics of the vineyards. Finally, through the creation of a web-based application, we have investigated the balance berry yield and wine quality. Using these tools further developed, vineyard owners should be able to predict the quality of the wine they intend to produce from their vineyards before the grapes are even harvested.

Fire detection is crucial for ensuring human safety and minimizing property damage. By utilizing advanced technologies, we can identify fires early, before they escalate. Autonomous fire detection systems are particularly vital in high-risk areas with minimal human presence. To address this need, we propose an automated fire-detection model based on YOLOv8. We modified the architecture of YOLOv8 and achieved impressive performances. Through comprehensive analysis of different YOLO models and their predecessors, we identified and addressed substantial gaps in this field. Also, by augmenting the framework with additional capabilities, we surpassed previous models, enabling real-time fire detection. Our proposed model achieves impressive performances, with 98% accuracy in fire detection and 97.8% in smoke detection. Moreover, recall rates for fire and smoke are 97.1% and 97.4%, respectively, with a mean average precision (mAP) accuracy of 99.1% for both cases. Finally, we applied an explainable artificial intelligence technique to interpret our proposed model and its results. This study lays the foundation for future research aimed at enhancing fire detection efficiency. The reliability of our modified model presents promising opportunities for future advancements, ensuring increased effectiveness and precision in efficient fire detection.

This paper provides a thorough examination of important themes critical to understanding and tackling modern cyber threats. It explores the revolutionary significance of AI and machine learning in cybersecurity, emphasizing their dynamic influence on defense systems. It also addresses how zero-trust designs have caused a paradigm change in network security, highlighting the significance of trust verification in current cybersecurity frameworks. Furthermore, the article investigates the integration of cybersecurity with corporate strategy, arguing for a collaborative approach to efficiently mitigating risks while minimizing operational impact. Furthermore, it investigates the idea of cyber resilience, highlighting an organization's ability to recover from assaults and adapt to changing threats. Furthermore, the paper discusses the issues faced by deepfakes in the synthetic media revolution and emphasizes the rising worry about data privacy in the digital era. Finally, it looks into the introduction of behavioral biometrics as a possible way to improve authentication and security methods. This paper contributes to the discussion of effective tactics for mitigating cyber threats in today's digital ecosystem by providing a thorough overview of these subjects.

Parkinson's disease (PD) is the second most common neurodegenerative disorder, entailing several motor-related symptoms that contribute to a reduced quality of life in affected subjects. Recent advances in wearable technologies and computing resources have shown great potential for the assessment of PD-related symptoms. However, the potential applications (e.g., early diagnosis, prognosis and monitoring) and key features of digital biomarkers for motor symptoms of PD (DB-MS-PD) have not been comprehensively studied. This study aims to provide a state-of-the-art review of current digital biomarker definitions for PD, focusing on the use of wearable devices. This review systematically examines research articles from 2012 to 2024, focusing on key features and recent technologies in PD research. A total of 22 studies were included and thoroughly analyzed. Results indicate that DB-MS-PD can accurately distinguish patients with PD (PwPD) from healthy controls (HC), assess disease severity or treatment response, and detect motor symptoms. Large sample sizes, proper validation, non-invasive devices, and ecological monitoring make DB-MS-PD promising for improving PD management. Challenges include sample and method heterogeneity and lack of public datasets. Future studies can leverage evidence of the current literature to provide more effective and ready-to-use digital tools for monitoring PD.

Water quality control employs techniques mostly targeting individual analytes; group detection is also practiced, but the choice of group methods is limited, which supports interest in developing such methods. We have examined an interaction of hypochlorite with a chlorine-containing heptamethine carbocyanine dye in the presence of 30 organic and inorganic model analytes that were found to induce diverse color changes in the system. The main supposed mechanisms are retardation of the dye oxidation with hypochlorite (presumably by scavenging chlorine radicals) and substitution of chlorine atom in the dye by the most nucleophilic analytes (amines, amino acids, proteins, DNA, phenol). The grass-green substitution product is more contrastingly visible against the dark-purple hypochlorite oxidation product of the dye than against the original emerald-green dye. The indicator reaction is monitored photographically for 10–40 min and the images are processed using principal component analysis (PCA) or linear discriminant analysis (LDA), allowing for data convolution for the complex color transitions. Nitrogen compounds are discriminated from the others, and more reactive analytes (tryptophan, cysteine, bovine serum albumin, and DNA) are detected in the presence of less reactive ones in natural water. The system is promising for the development of group assays for dissolved organic matter and discrimination of water samples.

In conventional Federated Learning (FL), clients work together to train a model managed by a central server, intending to speed up the learning process. However, this approach imposes significant computational and communication burdens on clients, particularly with complex models. Additionally, while FL strives to protect client privacy, the server's access to local and global models raises security concerns. To address these challenges, Split Learning (SL) separates the model into parts handled by the client and the server, though it suffers from inefficiencies due to sequential client participation. To overcome these issues, SplitFed Learning (SFL) was proposed, which combines the parallelism of FL with the model-splitting strategy of SL, enabling simultaneous training by multiple clients. Our main contribution is the theoretical analysis of SFL, which, for the first time, includes non-i.i.d. datasets, non-convex loss functions, and both full and partial client participation. We provide convergence proofs for a state-of-the-art SFL algorithm based on conventional convergence analysis assumptions for FL. Our results prove that we can recover the linear convergence rate of conventional FL for the SFL algorithm with the distinction that increasing the number of local steps or clients may not speed up the convergence in SFL.

Automating pruning tasks entails overcoming several challenges, encompassing not only robotic manipulation but also environment perception and detection. To achieve efficient pruning, robotic systems must accurately identify the correct cutting points. A possible method to define these points is to choose the cutting location based on the number of nodes present on the targeted cane. For this purpose, in grapevine pruning, it is required to correctly identify the nodes present on the primary canes of the grapevines. In this paper, a novel method of node detection in grapevines is proposed with four distinct state-of-the-art versions of the YOLO detection model: YOLOv7, YOLOv8, YOLOv9 and YOLOv10. These models were trained on a public dataset with images containing artificial backgrounds and afterwards validated on different cultivars of grapevines from two distinct Portuguese viticulture regions with cluttered backgrounds. This allowed to evaluate the robustness of the algorithms on the detection of nodes in diverse environments, compare the performance of the YOLO models used, as well as create a publicly available dataset of grapevines obtained in Portuguese vineyards for node detection. Overall, all used models were capable of achieving correct node detection on images of grapevines from the three distinct datasets. Considering the trade-off between accuracy and inference speed, the YOLOv7 model demonstrated to be the most robust in detecting nodes in 2D images of grapevines, achieving F1-Score values between 70 % and 86.5 % with inference times of around 89 ms for an input size of 1280×1280 px. Considering these results, this work contributes with an efficient approach for real-time node detection for further implementation on an autonomous robotic pruning system.