Abstract:

Londin et al. discovered miRNAs (LmiRNAs) whose properties had been little studied for unknown reasons. In this study, we examined fully complementary interactions of LmiRNAs with mRNAs of human genes. Using the MirTarget program, a significant number of target genes with unique properties of interaction with LmiRNAs were identified. Of 3707 LmiRNAs in the mRNAs of 75 target genes, the sites of fully complementary LmiRNA binding were located in the 5'UTR with a high free energy of interaction. In the mRNAs of 81 target genes, the fully complementary LmiRNA binding sites (BSs) were located in the CDS. Only seven LmiRNAs bound in the 3'UTRs of target genes. The KIFC3, PHF15, RPL15 and SNX11 genes were identified, encoding LmiRNA-5p and LmiRNA-3p, which bind to the mRNAs of these genes. The mRNA of most genes were bound by only one LmiRNA. The BMP8B, BMP8B, FGFRL1 and SDC3 genes were identified, whose mRNAs bound the pair ID00121.5p and ID02992.5p. These results expand our knowledge of LmiRNAs and allow us to recommend these LmiRNAs as diagnostic and therapeutic agents for various diseases.

Abstract:

The tectonic framework of the 2019 Mirpur seismic sequence was investigated using local seismic data. Moment tensor inversion was performed for the Mirpur mainshock (Mw=5.8) and its largest aftershock (Mw=4.7). The mainshock exhibited a low dip angle (~10°) and shallow focal depth, suggesting association with the Main Frontal Thrust (MFT). Contrary, the largest aftershock showed a comparatively higher dip angle, indicating deformation along a ramp or ramp-anticline structure. The stress regime was evaluated using moment tensor solutions from two locally inverted events, two Global CMT solutions and a published focal mechanism. The results indicated a maximum horizontal stress (SHmax) orientation of N07°E, consistent with the stress orientation of the 2005 Kashmir earthquake sequence and the regional compression driven by the Indian plate convergence. Aftershock relocation and spatial distribution patterns suggested post-seismic stress relaxation and possible activation of a higher-dip fault segment or ramp-anticline structure. Although a seismogenic depth of approximately 15 km was estimated, the concentration of seismicity around 10 km depth may involve the Main Himalayan Thrust (MHT). The Mirpur mainshock was triggered primarily by the movement along the MFT, whereas the subsequent aftershocks reflected post-seismic relaxation associated with the MHT and related ramp-anticline structures.

Abstract: This study proposes a unified gravitational theory framework based on discrete space element dynamics, grounded in two fundamental principles: spatial material conservation and global covariantity. The framework posits that spacetime consists of indivisible discrete space elements, where quantum virtual processes of matter generate new space elements through the consumption of conserved spatial materials. The resulting local density gradients constitute the microscopic essence of spacetime curvature. By eliminating superlative effects, this framework achieves self-consistency with general relativity under covariant constraints while fundamentally resolving four major physics puzzles: dark matter, dark energy, black hole singularities, and vacuum catastrophe.This paper first elucidates the core concept of "holistic covariant symmetry" and provides the ultimate explanation for symmetry breaking: symmetry breaking represents a local trade-off for achieving global covariance. Subsequently, it systematically expounds twelve key arguments of the framework, using the second-order discrete wave equation of complex fields as the sole foundational equation. Through rigorous step-by-step derivation, it rigorously establishes all fundamental laws of classical and quantum physics—including the Newtonian gravitational limit, mass-energy equivalence, the principle of the constancy of the speed of light, Maxwell's equations, Newton's three laws, Schrödinger's equation, and Dirac equation. The paper also clarifies the geometric origin of spin-1/2 and presents the geometric formula for the fine structure constant, demonstrating that all physical laws are theoretically derived rather than externally imposed.To address fundamental gaps in existing theories—including ambiguous definitions of spacetime structures, missing quantitative mapping for compactification, unclear Laplace approximation mechanisms, and undefined density-curvature relationships—this study introduces three core innovations: an asymmetric nanograding model, a compactified Landau free energy theory, and a third-order discrete Laplace operator with differential-geometric field mappings. All quantified parameters (error <1%, fit>0.95) are derived through rigorous first-principles calculations and constrained by observational data, eliminating any artificial adjustments. The research conducts cross-verification through eight modern geometric frameworks—including fiber bundles, complex geometry, and conformal geometry—unifying standard model constants as discrete spacetime invariants. Leveraging discrete compact manifolds and亏格 geometry, it achieves parameter-free numerical calculations of lepton mass ratios, derives the Friedmann equation with discrete geometry corrections, and provides a natural geometric explanation for the "cosmic lithium problem." The study ultimately delivers eight quantitative predictions verifiable by future high-energy physics and cosmological experiments, establishing a coherent, complete, and falsifiable new pathway toward unifying quantum gravity with the standard model.

Abstract: ABSTRACTIntimate partner violence is a serious public health issue with detrimental consequences on the victim's health. This study explores the perspectives of mental health professionals on their role in supporting women who seek help following the experience of intimate partner violence in the rural areas of the Eastern Cape, South Africa. The findings highlight the association between the experience of violence, poverty, lack of access to health services, and rural women's limited ability to escape abuse and recover. These structural factors contribute significantly to poor health outcomes, as this study found barriers preventing rural women from receiving counselling following the traumatic experience of abuse. Most notably, our findings uncover the impact of the shortage of shelters in rural areas, which often forces mothers to separate from their young children as they try to rebuild their lives after escaping the abuse. This unique insight distinguishes this study from previous work on this topic. The findings reveal a major gap in the public health response to intimate partner violence in rural areas of the Eastern Cape, particularly regarding the lack of accessible shelters for rural women. The study concludes that addressing the major shortcomings in the public health response to abuse of women in rural areas is critical to address poor health outcomes for women. We recommend increasing the quantity of shelters across the Eastern Cape and making them accessible to rural women and their children.

Abstract: Decentralised manufacturing is expanding as digitally controlled fabrication tools become accessible to SMEs, independent operators, and community workshops outside traditional factory settings, but the resulting heterogeneous, autonomously operated network introduces systemic uncertainty that no central authority governs. This paper proposes a systems-theoretic framework in which Free and Open Source Software (FOSS) governance acts as the structural interoperability layer of a distributed cyber-physical manufacturing system (CPS), and node-local digital twins --- each hosting a machine learning (ML) disturbance estimator --- provide local adaptive compensation without centralised data aggregation. A defining property of the architecture is automatic improvement propagation: learned corrections distribute via federated learning to structurally similar nodes without operator intervention, and the open, observable FOSS ecosystem enables advances in one fabrication modality to transfer to others through shared interface standards. The framework is applied analytically to three disturbance classes: regulatory restriction, technical process variability, and supply-chain disruption. Across cases, the analysis shows how open modular interfaces and local adaptation preserve functional continuity under perturbations that would more strongly affect centralised architectures. The contribution is a unified mathematical basis for robustness analysis in decentralised manufacturing CPS and a foundation for future simulation and empirical validation.

Abstract: This study adopts an agricultural archaeology perspective, integrating excavated remains, artifact genealogies, and pictorial materials to conduct a systematic investigation into the origins, evolution, and cultural significance of traditional Chinese tea-making techniques.By examining tea plant genetic remains dating back 6,000 years, tea-processing tools from the Western Han to Tang-Song periods, Ming-Qing purple clay tea ware, and representative tea paintings from the Tang, Song, Yuan, Ming, and Qing dynasties, this study analyzes the historical development of tea-making techniques, revealing their continuous evolution from the nascent stages of consumption to systematic development.The research demonstrates that archaeological evidence not only provides a solid foundation for chronological progression and technical analysis of traditional Chinese tea-making techniques and related customs, but also reflects the interactive relationship between technological innovation, the dissemination of tea customs, and social structures. Furthermore, archaeological material from Liao Dynasty tombs, Tibetan burial sites, and overseas shipwrecks indicates that tea customs exhibit remarkable cultural adaptability and influence in cross-regional exchanges and global dissemination.This paper argues that agricultural archaeology not only provides material evidence and methodological frameworks for studying traditional crafts but also offers new academic perspectives for understanding the diverse values of Chinese tea culture across temporal and spatial dimensions.

Abstract: Protected areas are often treated as internally homogeneous conservation units, yet their communities may be structured either as discrete modules or as continuous gradients shaped by environmental heterogeneity and human disturbance. Using camera-trap data from Liziping Nature Reserve, China, we examined the spatial organization of mammal and galliform bird communities and tested whether species-level environmental responses help explain community structure. From 148 camera-trap sites surveyed between July 2018 and June 2019, we obtained 4,065 independent detections and retained 15 species for analysis. We combined β-diversity decomposition, clustering, NMDS ordination, single-species occupancy models, clustering of environmental response coefficients, and Mantel tests. Community variation was dominated by turnover rather than nestedness, and clustering based on co-occurrence and relative activity patterns did not reveal well-separated discrete modules. Instead, NMDS indicated continuous variation along environmental gradients, with elevation and vegetation productivity as the strongest correlates. Occupancy models showed marked species-specific environmental responses, especially to elevation, habitat structure, and human disturbance, and β-based clustering identified two distinct environmental response groups. These results indicate that communities in Liziping are better characterized as continuous gradient structures than as discrete modules, and suggest that conservation should emphasize the maintenance of environmental heterogeneity, habitat continuity, and connectivity within mountain protected areas.

Abstract: This article provides a comprehensive critical analysis of Benjamin F. Jones's influential work on age and great invention, which documents a significant secular trend toward older ages at which inventors make breakthrough contributions. Drawing on data from Nobel Prize winners and great inventors across the twentieth century, Jones finds that the mean age at great invention increased by approximately six years over this period, attributing this shift to the expanding "burden of knowledge." This article examines Jones's theoretical framework, empirical methodology, and the broader implications of his findings for innovation policy and economic growth. While acknowledging the paper's substantial contributions, this analysis identifies important limitations—including concerns about measurement validity, alternative causal interpretations, and the generalizability of findings—and engages with contradictory evidence that complicates the burden of knowledge narrative. The article situates Jones's work within broader literatures spanning economics, psychology, and the sociology of science, ultimately arguing that while the burden of knowledge hypothesis offers a compelling partial explanation for observed trends, the phenomenon is likely more complex and contingent than the original framework suggests.

Abstract: The accurate prediction of feedback from user comments is essential yet challenging, often limited by the nuanced semantics that traditional Natural Language Processing and existing Large Language Model prompts struggle to capture. We propose the Hierarchical Feedback Reasoning Prompting (HFR-Prompt) framework to address this. HFR-Prompt guides Large Language Models through a multi-stage, logically progressive analysis comprising Initial Tendency Assessment, Fine-grained Feedback Type Identification, and Result Integration and Explanation Generation. Each successive stage builds upon the contextual understanding established by the previous one. Extensive experiments on a substantial dataset demonstrate that HFR-Prompt significantly outperforms strong LLM baselines and standard prompting techniques in terms of accuracy, Macro-F1 score, and crucial explanation consistency. While introducing a computational overhead, HFR-Prompt sets a new standard for interpretable and accurate comment feedback prediction, validating the efficacy of structured, hierarchical reasoning in complex LLM applications.

Abstract: Background: Goals of care discussions are essential communication skills in medical training that bridge patient values with clinical decision-making. Integrating palliative care principles into these conversations enables holistic, patient-centered care, yet medical trainees often lack structured preparation for these critical interactions. Objective: This narrative review examines how medical training can effectively integrate palliative care approaches into goals of care discussions through structured communication frameworks, interdisciplinary collaboration, and emerging innovations to promote patient-centered outcomes. Methods: Literature on evidence-based communication frameworks (SPIKES, REMAP, SUPER, Serious Illness Conversation Guide) was reviewed to identify training approaches. Clinical outcomes including patient satisfaction, hospice utilization, ICU transfers, and intervention intensity were examined. Educational barriers and facilitators—including communication training curricula, cultural competency, language considerations, and multidisciplinary team involvement—were evaluated. Emerging technologies supporting clinician education and practice were also assessed. Results: Training in structured communication frameworks improves patient-physician relationships, reduces patient anxiety, and increases family satisfaction. Early palliative care integration through effective discussions leads to increased hospice awareness and utilization while reducing burdensome interventions. Key educational facilitators include dedicated communication skills training, multidisciplinary team participation (including chaplains and palliative care specialists), and AI-assisted documentation tools that support learning while preserving humanistic clinician-patient interactions. Conclusions: Integrating palliative care principles into medical training for goals of care discussions is essential for developing patient-centered clinicians. Combining structured communication frameworks, interprofessional education, targeted skills training, and technological support creates a comprehensive educational approach that prepares trainees to elicit patient goals, create individualized care plans, and deliver holistic care that honors patient values.

Abstract: Microplastics have been known to kill fish and other microorganisms that feed on them in water bodies. The microplastics are also harmful to human beings when consumed directly or indirectly. This paper focuses on extracting features that can be used to build a model for identifying microplastics in images taken from open sewers that lead to the Indian Ocean. One thousand (1000) pictures were taken from selected points in Kilifi, Mombasa and Kwale counties in Kenya using a still picture camera. The pictures were then subjected to auto-cropping using a code written in python programming language. TensorFlow tool with openCV was used to capture the shape of the microplastics and annotate them by drawing bounding boxes. This was followed by application of Scale-Invariant Feature Transform (SIFT) algorithm to extract features from the images. The output of the process was a dataset of features for model building to identify microplastics in images. Further research can be conducted to extract more features using different algorithms and build models for identifying microplastics in images.

Abstract: After decades of in vivo isotope tracing, human solid tumors have not been shown to derive the majority of their carbon from circulating glucose. Despite this, glucose uptake by tumors continues to be widely interpreted as evidence of glucose dependence for growth. In contrast, mounting clinical and metabolic evidence indicates that glucose and glutamine are consumed primarily as regulatory and competitive substrates rather than as dominant carbon sources, with tumor biomass supplied largely by lactate, glutamine, and host-derived amino acids and lipids.Cachexia is commonly described as a secondary complication of advanced cancer, yet this metabolic behavior suggests it functions instead as a tumor-maintained systemic state that favors malignant survival at the expense of host tissues. By consuming glucose and glutamine at high rates, tumors restructure host metabolism, suppress immune function through substrate deprivation, and induce a catabolic shift that mobilizes host tissues as the tumor’s true nutrient reservoir. Dietary deprivation strategies therefore fail in solid tumors not because tumors adapt to starvation, but because restriction accelerates host metabolic collapse rather than depriving the tumor.Central to this argument is a newly proposed construct: homeostatic deception via dissociated catabolic ketosis, a tumor-orchestrated state in which physiological ketogenesis is genuinely present but decoupled from its normal protein-sparing function. Circulating ketones satisfy central energy-sensing mechanisms, silencing counter-regulatory alarms while unrestrained muscle proteolysis and lipolysis proceed. The resulting catabolic loop supplies tumors with substrates released from host tissues while the host’s regulatory systems interpret the state as normal adaptive fasting. Cachexia persists as long as the tumor driver remains active and reverses primarily when tumor burden and inflammatory signaling are controlled. A case of metastatic NSCLC, with photographic documentation, serves as the observational origin of this framework (Johnson CL, 2026, https://doi.org/10.5281/zenodo.18988466). This manuscript integrates metabolic tracing, immunometabolism, and clinical observation to propose a mechanistic hypothesis reframing cachexia as a tumor-maintained state. The framework identifies multiple targets for companion therapeutic intervention and explains the failure of diet-based strategies.

Abstract: This paper represents a further academic deepening and upgrading of the authors' 2019 publication A Hypothesis on the Spatial Motion Mode of Photons. It should be explicitly stated that this paper falls within the category of natural philosophical thought experiments—its core value lies in constructing a unified physical image of the nature of light through rigorous logical deduction, and proposing verifiable theoretical hypotheses and experimental schemes; the validity of all conclusions must ultimately be verified by rigorous and extensive scientific experiments before being incorporated into the theoretical system of physics. As a foundational concept of quantum mechanics, the wave-particle duality of light has been accompanied by profound philosophical perplexities and theoretical tensions since its proposal, becoming a core bottleneck in the integration of classical and quantum physics. This paper systematically sorts out the logical incompleteness in the current quantum interpretation system—including the self-negation of the complementarity concept, the problem of photon localization, the fundamental opposition between the statistical and non-statistical interpretations of the wave function, and the philosophical controversy over the Heisenberg Uncertainty Principle, revealing the inherent contradictions of the traditional wave-particle duality framework. On this basis, adopting classical physical images and the logic of reduction to absurdity, and based on six axioms and six preparatory propositions, this paper puts forward a natural philosophical hypothesis on the essence of photons: a photon is an energetic mass point with a diameter smaller than the Planck length, moving in a uniform spiral linear motion in space. The paper deduces the core characteristics such as velocity, frequency, and wavelength of the photon's uniform spiral linear motion, and designs three operable, repeatable, and quantifiable physical experimental schemes to provide specific paths for the empirical verification of the hypothesis. The research deduces that the angular momentum of photon spatial motion (excluding photon spin motion) is always the reduced Planck constant ℏ, the energy E=mc² is naturally unified with E=hν (the standard formula for wave energy), and the standard expression of the Heisenberg Uncertainty Principle ΔxΔpₓ≥ℏ/2 can be given a classical physical interpretation from the perspective of superposition of measurement deviations. This paper systematically responds to potential questions regarding the origin of photon particle nature, wave nature, and compatibility with relativity, arguing that the hypothesis provides a logically consistent and clearly visualized path for understanding the nature of light, builds a new natural philosophical framework for the integration of quantum and classical theories of light, and also offers a new thinking perspective for the paradigm shift in the study of the nature of light.

Abstract: Detectability of auditory signals in built environments is a critical issue in architectural acoustics, particularly in public spaces where notification sounds must be perceived reliably under background noise. This study investigated reaction times (RTs) to amplitude-modulated pure tones under silent, white noise, and bandpass-noise conditions. Twenty young and twenty elderly participants responded to 1- and 2-kHz tones with flat, gentle, and steep onset envelopes. To describe perceptual detection in physically interpretable terms, a time-integrated sound-exposure level model, LAE(t), was applied. RT was defined as the moment when cumulative acoustic energy exceeded a criterion value relative to the hearing threshold. In silent conditions, RTs were accurately predicted by LAE(t), with onset-envelope shape influencing early energy accumulation. In noise conditions, RTs increased systematically with spectral proximity between target and masker, consistent with auditory filter theory. When spectral separation exceeded approximately four ERB numbers, masking effects were minimal and RT approached silent-condition values. These findings demonstrate that perceptual detection timing is governed by cumulative acoustic energy and spectral masking rather than instantaneous sound pressure level. The LAE(t) model provides a detection-oriented metric that complements conventional room-acoustic parameters and may support evidence-based design of perceptually robust auditory signals in architectural environments.

Abstract: Textile dyeing industry is a significant contributor of complicated and extremely polluting wastewater. This wastewater has intermittent loads of chemical oxygen demand (COD), stains and other pollutants which puts dangerous effects on the sustainability of the environment and human beings in general. The traditional operation of wastewater treatment plants is reactive and rule-based to a large extent. These methods are ineffective in dealing with the non-linear dynamic character of the effluent of the textile business, resulting in low efficacy and recurring regulatory breach. To overcome these shortcomings, this paper will suggest a new hybrid architecture SAGE-GBTCN (Shock-Aware Gated Ensemble with Gradient Boosting and Temporal Correction Network) to be used in the effective prediction of wastewater pollution. This model combines a gradient boosting ensemble to produce baseline predictions and a parallel temporal network with a residual correction. A shock-sensitive gating system is used to dynamically modify the correction process to consider any sudden, non-stationary changes in the nature of the effluents. This design makes the model very useful in capturing the long-term trends as well as abrupt disruptions within textile wastewater. The suggested SAGE-GBTCN model was tested with the help of data on a full-scale wastewater treatment facility. The findings are shown to be more accurate in prediction and better resistant to abnormal operating condition. The model also demonstrates high possibilities to facilitate active and energy saving management of textile wastewater treatment processes, which will result in an R2 predictive value of 0.942 and a RMSE of 30.30 of COD. Although validated on full-scale industrial WWTP data, the proposed framework targets operational characteristics typical of textile effluent treatment plants, including batch-wise COD loading, abrupt shock events, and chemically driven variability.

Abstract: The intensification of ESG disclosure requirements under the Corporate Sustainability Reporting Directive (CSRD) and the International Sustainability Standards Board (ISSB) has increased the demand for artificial intelligence (AI) and data analytics to support large-scale sustainability reporting and verification. However, the existing academic literature remains fragmented across disciplinary domains, including natural language processing, machine learning, auditing, and regulatory compliance. This study addresses this gap through a PRISMA 2020-compliant systematic literature review of 45 peer-reviewed articles published between 2020 and 2025 and indexed in the Scopus database. The analysis combines bibliometric techniques using VOSviewer with qualitative thematic content analysis. The results reveal a rapidly expanding research field with a compound annual growth rate of 91.9%. Four major thematic dimensions emerge: (i) NLP and text mining for ESG disclosure analysis; (ii) machine learning applications for ESG scoring and corporate performance; (iii) AI-enabled ESG assurance, auditing, and governance; and (iv) regulatory frameworks and the digital transformation of sustainability reporting. The findings indicate that AI technologies are progressively transforming ESG disclosure from a predominantly narrative and self-reported practice into a data-driven and verifiable transparency system. These developments have important implications for regulators, corporate practitioners, assurance providers, and investors seeking to enhance the reliability and comparability of sustainability disclosures.

Abstract: The proliferation of 4D point cloud videos highlights their potential, but the high cost of obtaining large-scale annotated data severely limits supervised methods. Consequently, self-supervised learning (SSL) is vital for learning generalizable representations from unlabeled 4D data. While existing SSL frameworks, such as Uni4D, have made progress, they often struggle with fine-grained motion understanding in extremely dynamic scenes, maintaining robustness under severe occlusion, and developing explicit predictive capabilities. To address these, we propose Dynamic4D, a novel and robust self-supervised framework tailored for dynamic 4D point cloud understanding. Dynamic4D introduces an Adaptive Causal Temporal Attention (ACTA) mechanism in the encoder for explicit causal temporal modeling and dynamic region-focused learning. Its decoder employs Motion Prediction Tokens (MPT) to directly infer motion vectors for masked regions. A novel adaptive motion-sensitive masking strategy further enhances robustness by intelligently prioritizing high-dynamic zones. Our multi-objective pre-training strategy integrates a new Dynamic Perception Loss alongside geometric reconstruction and latent-space alignment. Extensive experiments on diverse challenging benchmarks demonstrate that Dynamic4D consistently achieves state-of-the-art performance. It substantially outperforms prior methods, validating its superior capacity to learn highly robust, generalizable, and motion-aware representations for complex dynamic 4D point cloud scenes.

Abstract: SARS-CoV-2 is a zoonotic virus with a proven ability to infect various animal species, including domestic cats. In the post-pandemic period of COVID 19, there are still limited data on the clinical course, shedding of infectious virus and diagnostic features in cats. The aim of this study was to investigate the spread of SARS-CoV-2 in cats in 2023, the clinical manifestations of the infection, the diagnostic algorithm including molecular detection of viral components, differential diagnosis of co-infection with FHV, FCV, Mycoplasma spp. and Chlamydia felis, serology and isolation of infectious SARS-CoV-2. The immunomodulatory therapy in animals with a standalone SARS-CoV-2 infection was applied. The study included oropharyngeal, conjunctival and nasal swab samples from 102 domestic cats with clinical signs. Of them, 20.6% (21/102) were positive for SARS-CoV-2, with 16.67% (17/102) of the cats showing various variants of co-infection with FHV, FCV, Mycoplasma spp. and Chlamydia felis. Four of the cats had a standalone SARS-CoV-2 with mild clinical manifestations that included serous discharges from the eyes, without change in the general condition. The virus was isolated from these samples. These four cats and their owners were positive for antibodies to the virus, and the owners were PCR-negative. The treatment of SARS-CoV-2 infection included the preparations Viusid, RX immunosuport, Vetomun and Lisymun. This is one of the first post-pandemic study covering FHV, FCV, Mycoplasma spp. and Chlamydia felis in domestic cats with SARS-CoV-2 infection and further expands on the essential main idea including the specified pathogens of interest.

Abstract: Intraoperative hypotension (IOH) is a critical complication during surgical procedures that can lead to severe adverse outcomes including myocardial injury, acute kidney injury, and increased mortality. Early prediction of hypotensive events remains a significant challenge in perioperative medicine. This study leverages the Medical Informatics Operating Room Vitals and Events Repository (MOVER) dataset, a comprehensive collection of intraoperative physiological signals and clinical events, to develop and evaluate machine learning models for predicting hypotensive events 5, 10, and 15 minutes before onset.The MOVER dataset contains high-frequency vital sign measurements including heart rate, blood pressure, oxygen saturation, and respiratory metrics from over 5,000 surgical procedures. Extensive preprocessing and feature engineering were performed to extract statistical, temporal, and interaction features across multiple time windows. Multiple machine learning algorithms were implemented and compared including XGBoost, Random Forest, Histogram-based Gradient Boosting (HGB), Support Vector Machines (SVM) with RBF kernel, Long Short-Term Memory (LSTM) networks, Multilayer Perceptron (MLP), and K-Nearest Neighbors (KNN).Experimental results demonstrate that XGBoost achieves the highest predictive performance with an accuracy of 94.2%, precision of 93.8%, recall of 94.5%, and AUC-ROC of 0.973 for 5-minute prediction windows. Performance remained strong for 10-minute (AUC-ROC = 0.942) and 15-minute (AUC-ROC = 0.908) predictions. Feature importance analysis revealed that mean arterial pressure (MAP) trends, heart rate variability, shock index, and time since last vasopressor administration were the most significant predictors. Error analysis identified borderline MAP values and rapid hemodynamic changes as primary sources of misclassification.The proposed models demonstrate strong potential for real-time clinical decision support systems to alert anesthesiologists of impending hypotensive events, enabling proactive interventions and improved patient outcomes. This research represents the first comprehensive comparison of multiple machine learning algorithms on the MOVER dataset for hypotension prediction, providing a foundation for future clinical implementation and prospective validation studies.

Abstract: The retention of Zillennial employees (born 1990–2005) presents significant challenges for contemporary organizations navigating competitive labor markets. This study investigates the relationships among perceived organizational support (POS), employee well-being (EWB), career development (CD), employee engagement (EE), and turnover intention (TI) within this workforce segment in Indonesia. Grounded in social exchange theory and complemented by conservation of resources theory, this research employed a quantitative cross-sectional survey design, collecting data from 360 Zillennial employees across multiple industries. Partial least squares structural equation modeling (PLS-SEM) tested the hypothesized relationships. Results indicate that POS (β = -0.285, p < 0.001) and CD (β = -0.198, p < 0.01) demonstrate significant negative direct effects on turnover intention, while EWB shows no significant direct relationship (β = -0.082, p > 0.05). All three antecedent variables significantly predicted employee engagement, which exhibited a strong negative relationship with turnover intention (β = -0.387, p < 0.001). Mediation analyses confirmed that employee engagement fully mediates the well-being–turnover relationship and partially mediates the effects of POS and CD. The model explained 64.8% of variance in turnover intention. These findings suggest that organizations seeking to retain Zillennial talent in Indonesia should prioritize organizational support systems, career development opportunities, and engagement-fostering initiatives. This study contributes to the literature by empirically examining these integrated relationships within an understudied demographic and cultural context, while acknowledging limitations inherent in cross-sectional, self-report designs.