Abstract: Public aircraft-engine certification data provide a reproducible basis for emissions modeling when proprietary combustor geometry, engine-cycle data, and detailed operating histories are unavailable. This study develops a physically constrained framework based on the International Civil Aviation Organization (ICAO) Aircraft Engine Emissions Databank for modeling gaseous emissions from civil aircraft gas-turbine engines over the landing-and-take-off (LTO) cycle. The analysis uses variables available directly from the ICAO Aircraft Engine Emissions Databank, together with derived LTO quantities, to construct combustor-aware reduced-order correlations for LTO-averaged emission indices of nitrogen oxides (NO_x), carbon monoxide (CO), and hydrocarbons (HC), denoted by EINO_xLTO, EICO_LTO, and EIHC_LTO, respectively. High-bypass-ratio (HBPR) turbofan engines are grouped by representative combustor technology, including conventional/single-annular combustor (Conventional/SAC), double-annular combustor (DAC), twin-annular premixing swirler (TAPS), Rolls–Royce TALON lean-burn combustor, low-emissions combustor (LEC), and Unknown categories. For each pollutant and combustor group, two-predictor quadratic response surfaces and power-law correlations are fitted using public engine-level predictors such as overall pressure ratio, bypass ratio, rated thrust, and total LTO fuel consumption. The results show that combustor-aware grouping substantially improves interpretability and that no single global predictor pair represents all pollutants or combustor technologies. For EINO_xLTO, overall pressure ratio appears in most selected predictor pairs, consistent with the pressure- and temperature-sensitive nature of NO_x formation. For EICO_LTO, robust group-specific correlations are obtained for several combustor classes, whereas EIHC_LTO is more sensitive to zero and near-zero values, making percentage-based errors and log-space power-law fits less reliable. The quadratic models generally provide stronger within-dataset descriptive fits, while the power-law models provide compact non-negative scaling relations when their errors are acceptable.The proposed framework is suitable for emissions-trend analysis, preliminary comparative assessment, and interpretation of public certification data, but it should not be used as a substitute for certification testing, detailed combustor simulation, or off-design mission-level prediction without additional validation.

Abstract: Recommender systems increasingly incorporate graph embeddings and graph neural networks to capture high-order relationships between users and items. However, the additional complexity of these approaches does not necessarily guarantee better recommendation quality than strong classical and latent-factor baselines. This study presents a reproducible comparison of six recommendation models representing four methodological families: Logistic Regression, Random Forest, Matrix Factorization with Bayesian Personalized Ranking, DeepWalk, node2vec, and LightGCN. The experiments were conducted on the MovieLens 1M dataset using a per-user temporal split. For each user, the most recent positive interaction was assigned to testing, the preceding interaction to validation, and all earlier positive interactions to training. All models were evaluated using identical candidate sets containing one held-out positive movie and 99 sampled unobserved movies. Performance was measured using Recall, Precision, Hit Rate, and NDCG at multiple cutoffs, complemented by bootstrap confidence intervals, paired statistical tests, computational-efficiency measurements, and analyses by user activity and movie popularity. Matrix Factorization achieved the best overall performance, reaching a Recall@10 of 0.7458 and an NDCG@10 of 0.4558, representing an approximately 56% improvement in NDCG@10 over Random Forest, the strongest classical baseline. LightGCN did not significantly outperform Logistic Regression and remained below Random Forest, despite its higher computational cost. DeepWalk and node2vec obtained similar and substantially lower aggregate results. Popularity-based analysis further revealed that classical models and LightGCN achieved substantially higher ranking effectiveness for popular movies, whereas Matrix Factorization maintained comparatively stronger performance for less-popular items. These findings demonstrate that model complexity alone is not a reliable indicator of recommendation effectiveness and highlight the importance of strong baselines, standardized evaluation, and reproducible experimental protocols.

Abstract: Nuclear Magnetic Resonance (NMR) spectrum, an invaluable tool in the frequency domain, is generated from the Fourier transformation of time-domain NMR signal. Previous research has utilized this method, applying the Fourier transform to data that represents the transient state solution of the improved Bloch NMR fluid flow equation. The current study developed a new resultant spectrum equation designed to enhance the accuracy of the NMR spectrum. Additionally, this research explores how the relaxation times of arterial, venous, and capillary blood affect the area beneath the resultant spectrum (A_r) produced by spinning blood protons. This study employed the Fourier transform to theoretically construct the frequency domain spectrum, while Laplace transform method and Heaviside expansion theorem were used to generate the NMR signals, which are the time-dependent solutions of the improved Bloch NMR fluid flow equations. For the purpose of data processing and visualization, MATLAB and Origin Pro software were employed. The arterial blood simulation results show corresponding values of A_r = 263.48746 Arads^-1m^-1, 261.59038 Arads^-1m^-1, 259.62945 Arads^-1m^-1, and 257.62471 Arads^-1m^-1 respectively at specific input values of arterial blood flow parameter T₁ = 1387 ms, 1419 ms, 1451 ms, and 1483 ms respectively. The simulation results produce values of A_r = 263.54293 Arads^-1m^-1, 281.53685 Arads^-1m^-1 , 299.23162 Arads^-1m^-1 , and 318.59282 Arads^-1m^-1 respectively for arterial blood when the input arterial blood flow parameter is set to T₂ =228 ms, 245 ms, 262 ms, 279 ms respectively. The simulation results produce values of A_r = 185.78128 Arads^-1m^-1, 185.81027 Arads^-1m^-1, 186.25025 Arads^-1m^-1, and 184.71834 Arads^-1m^-1 respectively for venous blood when the input venous blood flow parameter is set to T₁ = 1381 ms, 1461 ms, 1451 ms, and 1486 ms respectively. The simulation results produce values of A_r = 189.30669 Arads^-1m^-1 ,189.25119 Arads^-1m^-1, 220.72525 Arads^-1m^-1, and 236.86291 Arads^-1m^-1 respectively for venous blood when the input venous blood flow parameter is set to T₂ = 158 ms, 173 ms, 188 ms, and 203 ms. The results of simulation for capillary blood show corresponding values of A_r = 10.78129 Arads^-1m^-1, 6.453 Arads^-1m^-1, 5.85084 Arads^-1m^-1, and 5.63595 Arads^-1m^-1 at specific capillary blood flow parameter values of T₁ = 300 ms, 1200 ms, 2100 ms, and 3000 ms respectively. The simulation results produce values of A_r = 10.77763 Arads^-1m^-1, 102.02507 Arads^-1m^-1, 377.54099 Arads^-1m^-1 and 14345.86911 Arads^-1m^-1 respectively for capillary blood when the input capillary blood flow parameter is set to T₂ = 10 ms, 73 ms, 136 ms and 199 ms respectively. These findings indicate that the number of vascular compartments blood protons undergoing spin varies as a result of differences in relaxation times of arterial, venous, and capillary blood during circulation through blood vessels.

Abstract: Legal artificial intelligence is moving beyond answer generation into consequential workflows across legal research, drafting, compliance, litigation support, public legal services, and regulated practice. This shift exposes the limits of conventional evaluation: a plausible answer may still rely on inappropriate authority, outdated law, an inapplicable jurisdiction, or an unreviewable process. We present a structured mid-year review of developments in trustworthy legal reasoning made public between 1 January and 30 June 2026, integrating four coded datasets: 141 topic-relevant publications, 99 product launches or major updates, 53 curated public events, and 69 policy or regulatory records. In the publication sample, legal retrieval or retrieval-augmented generation appeared in 109 papers (77.3%), benchmark and evaluation research in 67 (47.5%), and legal agents or simulation in 30 (21.3%). Retrieval quality was addressed in 107 papers (75.9%), while temporal validity appeared in only 18 (12.8%) and uncertainty or refusal in nine (6.4%). Product activity was geographically concentrated: developers headquartered in the United States and the United Kingdom accounted for 68.7% of observed records, and major updates outnumbered new launches. Across the datasets, the field is moving toward workflow-level, retrieval-grounded, and agentic legal AI, but temporal and jurisdictional validity, actionable uncertainty, reproducible oversight, and contestability remain unevenly operationalised. Trustworthy legal reasoning is therefore not a property of a model alone, but an institutional achievement requiring authoritative sources, inspectable processes, accountable human roles, and effective routes for challenge and correction.

Abstract: Deterministic differential equations describe dynamical systems in idealized states, neglecting any random influences. Within biomathematical modeling, incorporating stochasticity requires a clear distinction between environmental (extrinsic) noise and demographic (intrinsic) noise. The latter framework assumes that temporal fluctuations arise strictly from the demographic dynamics of interacting populations rather than environmental variability. The literature thoroughly documents how demographic noise can be modeled and simulated as a stochastic process acting on individual members of a population, which yields discrete stochastic systems. For large population sizes, these discrete processes approximate continuous ones, leading to stochastic differential equations (SODEs). If random effects are omitted, these SODEs simplify back to standard ordinary differential equations (ODEs). Conversely, deducing how demographic noise impacts a natural system previously modeled by ODEs represents a major challenge. In this paper, we present an initial comparison of two distinct methodologies for reconstructing demographic noise by working backward from a deterministic, continuous differential system to its discrete stochastic process: the traditional Allen’s method and the backward approach recently introduced by Carletti and Banerjee (2019).

Abstract: Test-time refinement aims to improve generated programs through additional inference, but its value after an initial candidate has been produced remains unclear. We conduct a controlled evaluation of one-round Self-Refine and Self-Debug across seven models and three Python code-generation benchmarks. For each model and task, both methods refine the same initial candidate, allowing us to measure refinement gain without variation in initial generation. Self-Debug produced positive refinement gain in 16 of the 21 model-benchmark combinations and no change in the remaining five, with gains reaching +9.57 percentage points. In contrast, Self-Refine reduced correctness in 16 combinations, with losses of up to 8.07 percentage points, and produced positive gains in only four combinations. Repair-regression analysis showed that Self-Debug rarely damaged initially correct candidates, whereas regressions under Self-Refine frequently outweighed its repairs. Further analysis showed that execution feedback was beneficial only when models acted on diagnostic failures and produced effective revisions. Resource analysis showed that Self-Refine incurred greater token overhead despite generally reducing correctness, making its additional inference difficult to justify. Self-Debug provided a more favorable gain-overhead balance, but its monetary efficiency varied with model pricing, and its additional gains may offer limited value when initial generation is already sufficiently accurate. These results show that one-round refinement is not inherently beneficial and should be applied only when its expected gain justifies the additional computation and cost.

Abstract: This paper introduces the new three-dimensional (3D) fractional-order Zernike moments obtained by extending the radial component of classical Zernike polynomials to admit fractional exponents. This formulation provides a mathematically consistent extension of integer-order Zernike moments to fractional orders while preserving their orthogonality in the unit ball. A reformulation of the fractional Zernike functions and a recurrence relation are also presented for efficient computation of the fractional-order moments. Rotationally invariant descriptors are further derived to allow orientation-independent shape representation. Numerical experiments demonstrate that the proposed fractional-order moments improve the accuracy of shape reconstruction and representation compared to their integer-order counterparts, highlighting their effectiveness for 3D shape analysis.

Abstract: Road-safety authorities in different jurisdictions each hold accident records that, combined, could train stronger severity-prediction models, yet these records can be neither centralized, for privacy and governance reasons, nor directly merged, because jurisdictions encode severity under incompatible ontologies. This paper recasts that impasse as an information-fusion prob- lem and fuses heterogeneous, cross-jurisdictional crash sources into a single accident-severity model while every raw record stays at its source. Three components act together: model-level fusion under differential privacy, a reliability-weighted aggregation rule that trusts each source by its measured quality rather than its size, and a per-source centered logit-adjustment layer that reconciles mismatched label ontologies without double-correcting the shared class imbalance. This is, to the authors’ knowledge, the first approach to address multi-source fusion, local data retention, formal differential privacy, and label-ontology reconciliation jointly for accident-severity prediction. Across the United States (US Accidents) and the United King- dom (STATS19), harmonized to a common killed-or-seriously-injured target and evaluated over five seeds, the proposed methods significantly outperform a uniform private baseline at every privacy budget (p < 0.05); the private fused model reaches 0.599 balanced accuracy, within 0.025 of a centralized upper bound (0.624) while keeping data local, and retains the highest macro-F1 of all methods. Alignment significantly improves generalization to unseen jurisdictions for five of six held-out regions. The study is equally explicit about limits: alignment does not equalize sources, and a membership-inference attack reveals no measurable leakage for differential privacy to remove.

Abstract: Bluetongue virus diagnosis in Argentina, primarily via RT-qPCR, faces challenges due to the limited number of laboratories, many of which are located in regions where BTV is exotic and far from primary sample collection areas. Consequently, blood samples must be transported over considerable distances for analysis, posing significant logistical challenges due to the specific transport conditions needed. Therefore, the development of methodologies that optimize sample collection, preservation, and safe transport is of vital importance. The objective of this work was to standardize a sampling protocol for BTV using filter paper cards, to achieve adequate preservation and transport of blood samples for subsequent diagnosis. Hydration with TE at 37°C yielded the lowest Cq among tested conditions, with a detection limit of 2.75 10² gc/µL. We evaluated the stability of the sample on the cards by storing them at 25°C, 4°C, -20°C, and -80°C for 1, 7, 30, 60, 120, and 360 days. Stability was verified up to 360 days at all temperatures. Additionally, we confirmed the absence of infectious material in the card eluate. Finally, we conducted a preliminary field evaluation by collecting blood samples from cattle and sheep in tubes and on cards. According to the Cohen´s Kappa index, the agreement between both methods was almost perfect. The standardized methodology represents a significant advance in BTV diagnosis, as it will allow samples to be transported without refrigeration and with adequate biosafety conditions without losing diagnostic capacity.

Abstract: Freshwater quality is degrading globally, and regulatory monitoring remains largely retrospective, identifying non-compliance only after it occurs. This study identifies whether multi-year, regulatory-relevant compliance breaches can be forecast from sparse monthly monitoring records alone, and whether such forecasts improve on the assumption that next year resembles the current one. Using approximately two decades (2004–2024) of Land, Air, Water Aotearoa (LAWA) data from 497 South Island, New Zealand river sites, a single pooled gradient-boosted (LightGBM) classifier was trained to forecast Escherichia coli worst-band (Band E) non-compliance under the National Policy Statement for Freshwater Management at one-, two-, and three-year horizons, benchmarked against persistence, trend projection, and majority-class baselines under strictly temporal validation. The model discriminated breaches reliably (AUC 0.84–0.85) and exceeded persistence in balanced accuracy at all three horizons. Its principal value was early warning: among currently-compliant sites, it recovered roughly half of subsequent breaches, transitions that persistence cannot detect by construction, yielding a forward watchlist of 116 sites at risk of entering the worst band (Band E) by 2027, concentrated in pastoral catchments. Forecasting from monitoring data alone imposes an honest ceiling; scores are reported as risk rankings. The findings support a shift from reactive to anticipatory freshwater management.

Abstract: Green criminology refers to the study of harms on humanity, the environment and nonhuman species committed by powerful institutions and organisations as well as ordinary people. It addresses violations of environmental morality, environmental ethics, animal rights, nature rights, eco-justice and social justice. Analysing climate change crimes is particularly challenging. Few actors can cause great harm across the globe with many invisible victims. In addition, the International Court of Justice recently advised that nations that commit climate change crime can be held legally accountable if they fail to minimise severe harm to the climate system. Rising powers of far-right populists have come with increasing influence of climate change deniers over policy, and climate policies are denounced and aborted. In parallel, far-right politicians systematically use hate speech and hate crime as a tactic to polarise climate politics, mobilise ingroup followers, and delegitimise and silence pro-climate advocates, eventually harming democracy. This paper theorises on the links between climate change crime, hate crime and democracy crime and suggests a broadening of green criminology to better analyse crimes and victims of climate politics. Green criminology should not only address power in the political economy, but the ideologies and governance styles associated with politics.

Abstract: Fitness landscapes have long served as a conceptual tool in evolutionary biology. In recent 2 years, however, they have shifted from being viewed primarily as theoretical constructs 3 to being used to interpret experimental data and, increasingly, to guide evolutionary 4 outcomes in biomedical systems. In this review, we trace that progression across major 5 landscape formalisms, including Fisher’s geometric model, Wright’s adaptive landscape, 6 fitness seascapes, and empirical fitness landscapes. We aim to highlight the strengths 7 of each model and its potential for use in cancer research. Rather than treating these 8 frameworks as isolated models, we show how they collectively reflect the field’s maturation: 9 from describing adaptation in idealized settings to quantifying epistasis and evolutionary 10 accessibility, to interpreting experimental genotype-fitness data, and finally to motivating 11 strategies for steering evolution in contexts such as treatment resistance in cancer. We argue 12 that the central promise of the fitness landscape framework now lies not only in explaining 13 evolutionary dynamics, but also in enabling predictive interventions and overcoming 14 cancer evolution.

Abstract: This study evaluates the integrated sustainability potential of WtE technology scenarios across eight South African metropolitan municipalities, which collectively represent 40.14% of the national population. The assessment considers energy generation, greenhouse gas (GHG) emission reduction, landfill space savings, waste diversion, financial feasibility, and technology readiness, in alignment with the Waste-to-Energy Roadmap, National Waste Management Strategy, and Integrated Resource Plan. Seven scenarios were evaluated for the period 2028–2050, incorporating recovery rates that increase at 7–8-year intervals to ensure sustainable feedstock supply. These include S1 (Business-as-Usual/Landfilling), S2 (Landfilling with upgraded landfill gas recovery, LFG+), S3 (LFG+ with anaerobic digestion (AD)), S4 (LFG+ with AD and incineration), S5 (LFG+ with AD and pyrolysis), S6 (LFG+ with AD and gasification), and S7 (LFG+ with AD and plasma gasification). An Integrated Sustainability Performance Index (ISPI), combining 12 normalized energy, environmental, financial, and technological indicators, identified S4 as the most sustainable scenario, with an average score of 0.688 ± 0.044 and first-place ranking in all municipalities. S3 ranked second at 0.520 ± 0.085, while S7 ranked last. S2–S3 remained financially attractive, whereas advanced thermal technologies benefited municipalities with larger waste volumes. Integrated WtE systems support South Africa’s energy transition and sustainable waste management.

Abstract: Dual-parameter photonic sensors often report a single detection limit without specifying its statistical definition or calibration transferability. This fully computational study evaluates these issues in a coupled interface-mode multilayer for simultaneous refractive-index (RI) and temperature sensing. The two resonances occur at 1517 and 1651 nm, with loaded Q factors of 232 and 208 and RI sensitivities of 90.71 and 329.41 nm/RIU. Modal-overlap analysis links the sensitivity matrix to distinct thermal-to-index response ratios. Bounded nonlinear calibration yields held-out errors of 2.43 × 10⁻⁵ RIU and 0.155°C. At 1% false alarm and 95% detection, device-specific probability-of-detection limits are 4.36 × 10⁻⁵ RIU and 0.123°C; transferring one nominal calibration across devices worsens them by factors of 81 and 203 under the linewidth-limited repeatability model. A sensitivity-matched trivial control gives statistically equivalent yield within the declared margin, bounding the topological claim to the studied operating point. Structural audits further show that the sensing modes are cavity-selected, the nanolaminate provides no net measured performance benefit, and hyperbolicity is unobservable at normal incidence. The results support bias-aware detection and calibration-transfer analysis as more informative benchmarks than nominal sensitivity alone.

Abstract: We present a theoretical investigation of the attoclock method in laser-assisted photoemission (LAPE) processes. We analyze atomic photoionization driven by a short XUV pulse assisted by a circularly polarized strong infrared (IR) laser under accessible experimentally conditions. We calculate three-dimensional photoelectron momentum distributions (PMDs) in order to extract time delays from their angular information based on the attoclock principle. Furthermore, we explore different XUV parameters to determine the optimal achievable temporal precision for the laser-assisted attoclock (LAAC). The analyses performed within the the strong-field approximation (SFA) are further supported by improved Coulomb methods, including the Coulomb–Volkov approximation (CVA) and numerical solutions of the time-dependent Schrödinger equation (TDSE). Our work lays a theoretical foundation for understanding time-resolved electronic dynamics in cutting-edge ultrafast experiments of photonic collisions.

Abstract: Inverse airfoil design, recovering a geometry that produces a prescribed surface pressureor edge-velocity distribution, is recast here as a single determined nonlinear root-find ratherthan an objective-function search. Parameterising the surface with Class-Shape-Transformation(CST) coefficients that enter the geometry linearly makes the geometric sensitivity of the surfaceexact, constant and design-independent. It also makes geometric design constraints, among themleading-edge radius, trailing-edge thickness and inscribed area, linear algebraic rows rather thannonlinear predicates. Appending the CST coefficients as unknowns to a coupled viscous/inviscidNewton solver (mfoil) therefore converts constrained shape optimisation into constrained root-finding: one square system, no outer loop, no surrogate. The architecture is validated with afalsifiable self-consistency test in which a known CST coefficient vector is recovered from itsown self-generated target to ∥A−A∗∥_∞= 2.75 ×10⁻¹¹ in six Newton iterations. The recoveredgeometry reproduces the reference section’s fully released (natural-transition) aerodynamiccoefficients to ∆c_l = 3.4 ×10⁻¹². An ablation matrix identifies the primary uniqueness guard forthe resulting square system: sensitivity-optimal (QR-pivoted) selection of target stations, notinitial-guess quality, separates recovery of the true design from clean convergence to a spuriousbut residual-zeroing root. Measured against a competently-tuned nested Levenberg–Marquardtbaseline under two independent fair-paired controls, the monolithic architecture requires 3.1–8.1×fewer counted flow solves and 3.4–3.5×less wall-clock time. This is a real but modest reduction,not the two-to-three-orders-of-magnitude headline hypothesised a priori, and it comes withdeterminism, an exact analytic Jacobian for the constraint rows, and per-iteration failure-modediagnostics for which the nested baseline has no analogue. Generalisation is then evaluatedon two pre-registered panels. A 20-section NACA panel recovers all 18 generable sections to∥A−A∗∥_∞ ≤1.51 ×10⁻¹⁰; a 117-section panel drawn from the UIUC coordinate databaserecovers every one of its 83 converged sections to better than 10⁻⁴, while missing the pre-registered composite criterion on iteration count rather than on accuracy. Both panel outcomesare reported alongside the exclusions they rest on and the geometric bias those exclusions carry.The architecture is finally placed on a comparison table against MISES’s own modal inverse mode,the nearest prior CST-based inverse method, and the current generative and learned-surrogateinverse-design literature, on formulation class, cost, determinism and constraint-handling groundsrather than a single flow-solve number, since the methods are not commensurable on that axisalone.

Abstract: Newborn screening (NBS) is an essential public health intervention for early detection and secondary prevention of inherited disorders in asymptomatic infants at birth. Advances in this field and the accessibility of new screening methods and therapies in the last decade has surpassed insurmountably Malaysia's national NBS programme that has stagnated. Currently, our nationwide routine screening that is mandatory is limited to only two conditions: glucose-6-phosphate dehydrogenase deficiency and congenital hypothyroidism. While services such as universal newborn hearing and critical congenital heart disease screening are available in many settings, most treatable genetic and metabolic disorders remain unscreened. Consequently, preventable morbidity and mortality from inborn errors of immunity and metabolism, and genetic diseases persist. This position statement reviews the evidence for expanding the national NBS programme in Malaysia, arguing that a modernised, contemporary and pragmatic method is both clinically tenable and morally imperative. By embracing genomic-era advancements, Malaysia can safeguard children's right to an open future, fulfil intergenerational justice, and ensure no infant is left without the opportunity for life-saving care.

Abstract: Electric vehicles (EVs) are increasingly recognised as a key component of sustainable mobility transitions. While technological advancements and supportive policies have significantly expanded EV adoption, the relatively high purchase cost of conventional EVs remains a barrier for many consumers. In response, a new generation of frugal EVs has emerged, emphasizing affordability, resource efficiency, and functional adequacy while maintaining essential mobility requirements. Despite growing commercial interest, the factors influencing the acceptance and affordability of frugal EVs remain underexplored in the academic literature. This paper addresses this gap through a structured literature review using the SALSA (Search, Appraisal, Synthesis, and Analysis) framework. A total of 116 studies were selected and analysed to identify the key factors influencing user acceptance and affordability in the context of frugal EVs. The reviewed evidence was synthesised into four principal dimensions: demographic, situational, contextual, and psychological factors. The findings indicate that affordability, low operating costs, compact urban mobility, modular vehicle design, and battery-swapping technologies are among the most important drivers of frugal EV adoption. In contrast, safety concerns, limited driving range, charging infrastructure constraints, weak resale value, regulatory uncertainty, and perceptions of lower quality remain significant barriers to market acceptance. The review further suggests that the adoption dynamics of frugal EVs differ from those of conventional EVs, particularly due to their emphasis on cost-performance trade-offs and urban mobility applications. From a smart-city perspective, frugal EVs have the potential to support more inclusive and sustainable urban transportation systems by expanding access to low-carbon mobility for cost-sensitive users and underserved market segments. The study contributes to the emerging literature on affordable electric mobility by providing a dedicated framework for understanding frugal EV acceptance and affordability. It concludes with recommendations for policymakers, manufacturers, infrastructure providers, and service operators aimed at supporting the wider adoption and distribution of frugal EVs.

Abstract: Facial aging is not merely a linear degeneration of the epidermis, but rather a complex, multilevel biomechanical cascade involving skeletal remodeling, dynamic redistribution of fat compartment volumes, and degradation of the skin matrix [1]. Modern anatomical studies have confirmed that subcutaneous fat is precisely divided into superficial and deep fat compartments with distinct boundaries and different aging trajectories, the atrophy of the deep fat compartments, combined with the displacement of the superficial fat compartments, leads to a significant reduction in facial three-dimensional volume [3]. At the same time, the functional decline of facial supporting ligaments is closely related to changes in the extracellular matrix's molecular structure, this reduction in tissue stiffness and tensile strength provides the pathophysiological basis for the transformation of dynamic wrinkles into static wrinkles [17]. In the field of computer vision, facial aging analysis primarily focuses on either inferring an individual’s physiological age from image sequences or enabling cross-age identity recognition, these technologies have broad application value in public security surveillance and digital forensics [7]. With the introduction of deep learning technologies, convolutional neural networks (CNNs) have become the mainstream analytical method in this field due to their exceptional feature extraction capabilities [8]. Addressing the significant variability in individual aging patterns, research has proposed models that incorporate attention mechanisms to achieve precise modeling of aging trajectories [11]. Generative adversarial networks (GANs), through age-conditional generation techniques, have successfully achieved age transformation while preserving identity features, effectively resolving the challenge of maintaining gender and racial consistency that traditional methods struggle with [14]. In age estimation tasks, ordinal regression and label distribution learning have further improved prediction accuracy by capturing the temporal order of age labels [13]. The core paradigm of cross-age face recognition is feature disentanglement, which aims to decompose facial representations into identity-inherent and age-related components, thereby minimizing the interference of age-related changes on identity recognition performance [12]. Despite significant progress, the field still faces severe challenges, such as insufficient data quality and diversity, as well as vast differences in individual aging patterns [7]. The “black-box” nature of deep neural networks results in a lack of sufficient theoretical explanation for the relationship between the learned aging representations and actual biological mechanisms, limiting their in-depth application in the biomedical field [15]. The scarcity of longitudinal paired data severely limits the models’ generalization ability, while the insufficient scale and diversity of mainstream datasets make overfitting a common issue [8]. Although the emerging line-scan confocal optical coherence tomography (OCT) technology can provide three-dimensional microscopic insights into the dermal fiber network, it still faces technical bottlenecks in aligning and standardizing multi-source data during clinical translation [16]. Future research must move beyond a mere race for performance and shift toward an in-depth exploration of the intrinsic structure and causal interpretability of aging characteristics. By integrating multi-source data, such as genomics, and establishing a new evaluation paradigm to assess the biological plausibility of models, thereby bridging the gap between deep network features and actual biological mechanisms of aging [37].

Abstract: Aberrant systemic IgG responses to the gut microbiota may be implicated in the pathogenesis of ulcerative colitis (UC). In this study, we investigated IgG-targeted bacteria and their predicted functional characteristics in UC. Using fecal samples and autologous sera from 37 patients with UC and 4 healthy controls, we profiled IgG-targeted bacteria under in vitro conditions using 16S rRNA gene sequencing. To correct for background bacterial abundance, we calculated the IgG+ Probability Score (IPS), which integrates the sample-specific IgG-binding rate to isolate taxa explicitly targeted by IgG. Systemic IgG responses were selectively targeted at specific bacterial taxa rather than uniformly directed toward the entire gut microbiota. Active UC and elevated systemic inflammatory markers were associated with increased IPS in oral-associated commensals, including Granulicatella, as well as mucin-degrading taxa. Functional profiling indicated the enrichment of pathways related to host-derived mucin degradation, capsular polysaccharide biosynthesis, and amino acid biosynthesis. Furthermore, Random Forest machine learning models utilizing optimized IPS features achieved an area under the curve of 0.817 for predicting UC exacerbation and 0.843 for predicting C-reactive protein elevation. In conclusion, systemic IgG responses in UC were selectively directed toward specific bacterial taxa associated with disease activity and systemic inflammation. Analyzing the IgG-targeted microbiota via IPS profiling deepens our understanding of host-microbiota interactions in UC and provides a valuable framework for evaluating immune-targeted bacteria that fluctuate across different disease phases.