Search Results (31,056)

Background/Objectives: Population differences in immunoglobulin (IG) variation offer a starting point for studying inherited antibody diversity. We screened reference-based single-nucleotide variants for population differentiation and linked the resulting candidates to expression and trait annotations. Methods: The primary analysis included 2488 participants from five 1000 Genomes Project groups after pruning documented pedigree relationships and incorporated reference-risk filtering, genotype-quality sensitivity, two FST estimators, linkage disequilibrium and resampling. A separate descriptive extension included 255 Simons Genome Diversity Project participants but did not constitute independent replication. Results: The four-estimate screening rule retained 231 conditional candidates: 49 in IGH and 182 in IGL, with none in IGK. Of these, 177 matched GTEx significant-eQTL catalogs, and seven coordinates overlapped 18 GWAS Catalog records. At rs2856876, the observed alternate-allele frequency was 70.04% in East Asian and 13.65% in European samples; annotations included an IGLV4-69 expression association and a protein-level trait with the same gene label. This correspondence identifies a question for molecular follow-up, not cross-assay validation. High-FST candidates were not consistently more likely to appear in GTEx catalogs than variants in supported local backgrounds. Local assembly–call disagreement, LD dependence and SGDP population composition further constrained interpretation. Conclusions: The resource connects conditional population differences with specific expression and molecular-trait records for follow-up studies of antibody variation. Full article

Background: Persistent postural-perceptual dizziness (PPPD) is a chronic functional neuro-otologic disorder marked by persistent dizziness, unsteadiness, and visual dependence. Its electrophysiological mechanisms remain unclear. This study examined resting-state EEG spectral power and functional connectivity in PPPD and their associations with postural stability and clinical symptoms. Methods: Forty patients with PPPD and 40 age- and sex-matched healthy controls were studied with an eyes-closed resting-state EEG recording. Static postural stability was measured with the Pro-Kin system. Spectral power was calculated for delta, theta, alpha, beta, and gamma bands. Functional connectivity was assessed using the weighted phase lag index. Results: Static posturography showed impaired postural stability in patients with PPPD compared with healthy controls. Spectral power analysis showed increased theta-band relative power over the frontal and left parieto–occipital regions. Connectivity analysis showed increased prefrontal–temporal and decreased temporal–occipital connectivity in the theta band, increased fronto–parieto–occipital and decreased parietal–temporal connectivity in the alpha band, and decreased cerebellar and temporal, central connectivity in the gamma band. Conclusions: PPPD was associated with frequency-specific alterations in resting-state cortical oscillations and functional connectivity involving cognitive–affective, visual, and sensorimotor networks. Resting-state EEG may help characterize network-level abnormalities in PPPD. Full article

Accurate remaining useful life (RUL) prediction of lithium-ion batteries is essential for ensuring the reliability and safety of battery management systems. However, conventional electrical and thermal signals are sensitive to operating conditions, while linear feature extraction methods may not adequately characterize the nonlinear electrochemical aging information embedded in electrochemical impedance spectroscopy (EIS). Moreover, the unequal degradation relevance of different impedance-frequency regions is rarely considered. To address these limitations, this paper proposes a frequency-weighted EIS manifold-learning framework for battery RUL prediction. A band-level frequency-weighting strategy is first introduced to incorporate degradation-related frequency priors, after which kernel principal component analysis (KPCA) is employed to construct compact nonlinear representations of EIS evolution, followed by support vector regression (SVR) with hyperparameter optimization. Linear bias correction and moving-average smoothing are further incorporated to improve prediction consistency. Experiments on seven batteries within the investigated LR2032 dataset, including one chronological test and six held-out-cell tests, demonstrate that the proposed framework consistently outperforms the baseline models, achieving an average MSE of 59.434 cycles² and an average $R^2$ of 0.989. Full article

Background: Gastrointestinal (GI) involvement is a clinically relevant but still incompletely characterized manifestation of pediatric Behcet’s disease (PedBD). This systematic review provides a focused analysis of GI involvement in PedBD (addressing its frequency, clinical symptomatology, anatomical/endoscopic features, and therapeutic management) as a dedicated sub-analysis of a broader systematic review. Methods: A systematic search of MEDLINE, EMBASE, CINAHL, CENTRAL, and ClinicalTrials.gov was conducted, following PRISMA 2020 guidelines. Observational studies reporting GI involvement in patients with PedBD (disease onset before 16 years of age) were included, and data on frequency, symptoms, anatomical/endoscopic findings, and treatment of GI-PedBD were extracted and synthesized. Results: Thirty studies were included. Most studies were not focused on GI manifestations in PedBD as a (primary) study aim. The frequency of GI involvement ranged widely across the selected articles with a median of approximately 15% among studies with ≥50 pediatric BD patients. Abdominal pain and diarrhea were the most consistently reported symptoms. Endoscopic data indicated frequent involvement of the ileocecal region and terminal ileum, with mucosal ulcers and erosions as the predominant findings. Therapeutic data specific to GI-PedBD were limited to two studies only. As an appendix to the systematic review in the discussion, relevant case reports were described: anti-TNF agents, particularly adalimumab, showed favorable outcomes in refractory cases. Conclusions: GI involvement in PedBD is not negligible and can be severe. Its clinical and endoscopic characterization remains incomplete due to substantial heterogeneity in reporting practices across studies. Anti-TNF agents may represent an effective therapeutic option, especially for refractory GI-PedBD. Prospective and multicenter studies with standardized GI assessment protocols are needed to better characterize this clinically relevant subgroup of PedBD patients. Full article

The long-term reliability of selective laser melting (SLM) equipment is constrained by the latent degradation and coupled faults of multiple underlying hardware systems under continuous high-load operations. Transitioning from passive defect detection to proactive prognostics and health management (PHM) is crucial to overcome bottlenecks in industrial applications. Accurately inferring future multi-system hardware states faces three major challenges: feature alignment difficulties under variable-length printing cycles, graph topology collapse under strong industrial noise, and dimensional conflicts along with error cascading divergence during the joint optimization of microscopic physical trajectories (continuous regression) and macroscopic system anomalies (discrete classification) in end-to-end prediction. To address these issues, this paper proposes a decoupled spatiotemporal graph autoregressive network (DS-GAN). First, a multi-scale feature pooling and degradation gating injection mechanism is constructed to align highly variable-length high-frequency sequences and adaptively integrate macroscopic health priors. This approach achieves feature decoupling under physical boundary constraints. Second, a restricted residual graph evolution mechanism is introduced to regulate dynamic coupling drift based on static physical topologies, effectively suppressing feature divergence in the spatial dimension. Finally, a heterogeneous multi-task autoregressive decoder based on homoscedastic uncertainty is designed; this decoder helps reduce the impact of accumulated errors along the time axis during multi-step forecasting. Long-sequence forward inference on a real SLM continuous printing dataset demonstrates that DS-GAN achieves an overall accuracy of 97.48% for multi-system anomalies while strictly limiting the global false alarm rate to 1.85%. Furthermore, quantitative results reveal the physical inertia mechanism within the prediction horizon, demonstrating that the model maintains high fidelity for physical trajectories with a Macro-RMSE of 0.062, even under a maximum predictive horizon of 20 s. This study provides a reliable theoretical and engineering framework for dynamic coupling correlation analysis and proactive fault warning of multi-systems in complex industrial equipment. Full article

Background/Objectives: Stable angina is a significant health issue in Egypt, often showing poor responses to standard treatments. This study examines the effectiveness and safety of early combination therapy with trimetazidine and either a β-blocker or a calcium channel blocker (CCB) among Egyptian patients with stable angina. Methods: This sub-analysis of the Egyptian cohort from the COMBINE Angina study included patients with stable angina (CCS class II–III) who remained symptomatic after 2–4 weeks of β-blocker or CCB monotherapy. Patients received trimetazidine 35 mg modified-release twice daily as an add-on treatment for 4 months. The primary outcome was the Seattle Angina Questionnaire (SAQ-7) summary score, with secondary outcomes including SAQ-7 domain scores, CCS class, frequency of angina attacks, nitrate use, treatment satisfaction, adherence, and safety. Results: In the global COMBINE study involving 578 patients, the Egyptian cohort had 138 participants with an average age of 58.3 years. The SAQ-7 summary score improved from 42.2 at baseline to 76.8 at month 4 (mean change 34.6 points, 95% CI 31.3 to 38.0; p < 0.0001). All 138 patients contributed data at every visit for the SAQ-7 summary score. Patients in CCS class I increased from 0% to 56.5%, weekly angina attacks fell from 5.0 to 1.7 (p < 0.0001), and short-acting nitrate use decreased from 2.9 to 1.0 doses per week (p < 0.0001). Patient satisfaction was about 75%, and physician satisfaction reached 83% by month 4. Medication adherence also improved (MARS-5 score: 20.8 to 22.7), and two non-serious, treatment-related adverse events (nausea and somnolence) were reported in a single patient (0.7%), with no serious or fatal events. Conclusions: In Egyptian patients with stable angina, implementing an early combination strategy incorporating trimetazidine with first-line β-blocker or CCB therapy was associated with clinically meaningful improvements in angina symptoms, functional status, and quality of life, with good tolerability and high satisfaction. As the study was observational and single-arm, these findings are provisional and require confirmation in further long-term controlled studies. Full article

Auditory evoked potentials (AEPs) comprise a family of electrophysiological responses that follow acoustic stimulation from the auditory nerve and brainstem to thalamocortical and association networks. Their interpretive value lies in the combination of millisecond temporal resolution and hierarchical sensitivity: early responses mainly index the fidelity and timing of ascending sensory conduction, whereas later responses increasingly depend on cortical context, attention, prediction, and behavioral relevance. This review develops a state-aware framework for interpreting AEPs during wakefulness, natural sleep, pharmacologic sedation, general anesthesia, and impaired consciousness. We first describe the ascending auditory pathway and relate approximate latency ranges to distributed neural generators. We then examine how attention and task engagement modify early cortical processing; how non-rapid eye movement and rapid eye movement sleep preserve some forms of acoustic analysis while reorganizing or suppressing others; and how anesthetic effects vary with response latency, drug concentration, and cortical hierarchy. The review also distinguishes transient event-related potentials from steady-state auditory evoked potentials (SSAEPs), and explains why phase-locked activity can be recovered by trial averaging whereas non-phase-locked induced activity requires single-trial time-frequency analysis. Separate sections address cross-species translation; click, chirp, oddball, steady-state, and naturalistic speech paradigms; as well as clinical applications in hearing assessment, intraoperative monitoring, anesthesia, and disorders of consciousness. The central conclusion is that an AEP is interpretable only when its generator, stimulus, analysis domain, recording conditions, and cognitive state are considered together. Full article

Welded structures are widely used in ships and marine equipment. The ultrasonic longitudinal critically refracted (LCR) wave method is a reliable technique for measuring residual stress in welded structures. Conventional ultrasonic measurement systems have high power consumption and are difficult to operate under the limited power supply conditions of marine environments. Reducing the excitation voltage can lower power consumption, but as the excitation voltage decreases, ultrasonic echo energy weakens and noise interference increases, thereby affecting the accuracy of stress measurement. This paper proposes a complexity-mutation-based adaptive singular spectrum analysis–variational mode decomposition algorithm (CM-SSA-VMD). The algorithm constructs a complexity index using the spectral centroid, waveform roughness, and zero-crossing rate. By identifying abrupt changes in complexity between adjacent singular spectrum analysis (SSA) components, it adaptively determines which components to retain for reconstruction. Then, variational mode decomposition (VMD) is used to further separate the residual high-frequency noise. Finally, the ultrasonic time of flight (TOF) is estimated using the cross-correlation algorithm, and the stress is calculated. The stress measurement accuracy of the proposed algorithm was evaluated at different excitation voltages: 5.2 V, 3.3 V, 1.2 V, and 0.5 V. The results show that when the excitation voltage drops to 0.5 V, after being processed by CM-SSA-VMD, the average relative error of stress measurement remains below 10%, while the average relative errors of a new adaptive denoising method, Grey Wolf Optimization–Variational Mode Decomposition–Wavelet Transform (GWO-VMD-WT), traditional FIR filtering and VMD are approximately 14%, 18% and 16% respectively. Compared with GWO-VMD-WT, FIR filtering and VMD, the measurement accuracy of this method is improved by 33.35%, 46.16% and 41.82% respectively. The CM-SSA-VMD algorithm can effectively suppress noise in ultrasonic signals at low excitation voltages and improve the reliability of stress monitoring. It provides a feasible method for low-power ultrasonic residual stress monitoring in marine engineering. Full article

Human factors remain a leading cause of occupational accidents and fatalities in the construction industry. Although human factors are a major contributor to construction accidents, their relative impact across different construction subsectors has not been fully quantified. This study presents a comparative analysis of accident and fatality data from the U.S. Occupational Safety and Health Administration (OSHA) for three major construction subsectors, NAICS 236 (Construction of Buildings), 237 (Heavy and Civil Engineering Construction), and 238 (Specialty Trade Contractors) from January 2010 through June 2025. Seventy human-factor-related keywords were categorized into eight categories, and accident counts, fatality counts, and fatality rates, defined as the proportion of investigated cases that were fatal, were computed. Twenty-four high-frequency keywords (≥60 investigated accidents) were further analyzed across the three subsectors. Descriptive methods, including risk matrices, Pareto analysis, and category-level fatality rate comparisons, were applied to identify the factors contributing most to fatalities. Differences between subsectors were tested using permutation-based chi-square tests with Benjamini–Hochberg correction for multiple comparisons. Results indicate substantial variation in both frequency and severity among keywords. At the category level, cognitive or perceptual errors, procedural violations and unsafe acts, and judgment and awareness failures in equipment use showed significantly higher fatality proportions in Heavy and Civil Engineering Construction than in the other two subsectors, whereas at the keyword level only lost balance and inattention differed significantly once the correction was applied. The findings provide a data-driven basis for prioritizing targeted safety interventions and tailoring human factor mitigation strategies to subsector-specific risks. Full article

Entropy estimation from short samples recurs in symmetric cryptography, where the reference distribution is uniform by design and no single estimator in the considered classical comparison set minimizes mean squared error (MSE) across the full range of ratios $n/k$. We introduce the adaptive sample-conditional entropy diagnostic (ASED), in which a compact network trained offline maps a frequency-of-frequencies descriptor of the sample to convex mixture weights over six classical estimators, at a cost of $O(n+k)$. We prove an oracle inequality bounding the excess risk of such a mixture by the $L_1$ error of its weights and fixed-alphabet consistency for every simplex-valued weighting rule, independently of the distribution used to train the weights. Under the uniform-null protocol, ASED attains an integrated MSE of $1.7\times {10}^{-3}$ for bytes, compared with $3.8\times {10}^{-2}$ for James–Stein shrinkage, a reduction that depends materially on the aggregation scheme: $95.4\%$ for summed or averaged MSE across sample sizes versus $24\%$ for the mean of per-size ratios. Both figures are reported together throughout, and neither is presented as the headline; the entire advantage is confined to the undersampled regime $n<k$, the two estimators being indistinguishable for $n\ge k$. A locked train/validate/test evaluation with an independently written implementation reproduces the reduction at $95.1\%$. Because a constant output achieves zero error by construction here, we add further checks: in a post hoc analysis, ASED is non-inferior to SHR in detection power at a $0.02$ margin, both pointwise and simultaneously, whereas a constant control has none, and the advantage persists on held-out configurations. The $0.02$ margin was selected after inspecting the power estimates, and at $n=8$ and $n=16$, it is smaller than the variation induced by tie handling at the empirical critical value, so the non-inferiority conclusion is confirmatory only for $n\ge 32$. At $n\le 16$, the two statistics have identical null distributions up to monotone relabeling, and no test of size $0.05$ exists, so no power difference is identifiable in either direction, and the comparison is withdrawn there. For strongly non-uniform sources, the ordering reverses, and NSB dominates, delimiting ASED as a uniformity diagnostic rather than a general-purpose entropy estimator. The design choices behind ASED, the feature map, architecture, and loss weighting were made while observing results on this same evaluation protocol, so no independent model-selection split separates development from the assessment reported here. That pipeline is accordingly designated exploratory, and the locked evaluation is confirmatory. The contribution is stated as sample-conditional convex aggregation of classical estimators rather than as the particular network realizing it: a five-feature variant and a gradient-boosted surrogate perform comparably, with no paired interval separating the three. Claims are restricted to short i.i.d. samples and marginal Shannon-entropy diagnostics under a uniform null; min-entropy, unpredictability, dependence, and randomness certification are out of scope. Full article

To characterize pesticide residues and age-specific dietary exposure risks in apples from Xinjiang’s main production areas, 120 samples were collected from Aksu, Ili, and Kashgar in October 2025 and screened for 82 pesticides using GC–MS/MS, LC–MS/MS, and GC–ECD. Chronic, acute, and cumulative dietary exposure was subsequently assessed. Eight pesticides were detected, all below their corresponding maximum residue limits (MRLs). Acetamiprid had the highest detection frequency (28.3%), yet pyridaben was the leading contributor to chronic hazard index (HI_c) under both lower-bound (LB) and upper-bound (UB) scenarios. Carbendazim, acetamiprid, and pyridaben were the major contributors to summed acute hazard index (HI_a), demonstrating a mismatch between detection frequency and risk contribution. The sample-level chronic hazard index was significantly higher in the multi-pesticide co-occurrence group than in the single-residue group (p < 0.001). Toddlers showed the highest cumulative exposure, with chronic hazard indices of 0.179% (LB) and 2.209% (UB), and an acute hazard index of 33.759%. Sensitivity analysis revealed that adopting the recently updated EFSA toxicological reference values for acetamiprid caused acute hazard quotients to exceed the 100% screening threshold in toddlers and children, highlighting how the source and currency of toxicological reference values can affect risk assessment conclusions. Under the primary analysis reference values, dietary exposure risks were low; however, risk prioritization should account not only for detection frequency and MRL compliance but also for actual exposure levels, co-occurrence patterns, and updates to toxicological reference values. Full article

Background/Objectives: Non-suicidal self-injury (NSSI) is an important mental health concern among college students, yet evidence regarding behavioral and psychological correlates of NSSI among male students remains limited. This study examined the cross-sectional associations among physical activity (PA), psychological resilience (PR), and NSSI in male college students and explored heterogeneity in joint PA–PR patterns. Methods: A cross-sectional survey was conducted among 2567 male college students from one university in Jiangxi Province, China. PA, PR, and NSSI were assessed using the Physical Activity Rating Scale-3, the 10-item Connor–Davidson Resilience Scale, and 10 behavioral-frequency items from the Ottawa Self-Injury Inventory, respectively. Pearson correlations, covariate-adjusted regression analyses, bootstrap analysis of the statistical indirect association, sensitivity analyses, and latent profile analysis (LPA) were performed. Results: PA was positively associated with PR (r = 0.627, p < 0.001), whereas PA and PR were negatively associated with NSSI (r = −0.259 and −0.292, respectively; both p < 0.001). After adjustment for all measured demographic characteristics, PR statistically accounted for part of the negative association between PA and NSSI (indirect effect = −0.133, 95% bootstrap CI [−0.164, −0.102]). The NSSI model explained 9.8% of the variance. Past-year NSSI was reported by 35.41% of participants. LPA identified two predominantly level-based profiles: Low PA–Low PR and High PA–High PR. Participants in the Low PA–Low PR profile reported significantly higher NSSI scores than those in the High PA–High PR profile. Conclusions: Higher PA and PR were associated with lower NSSI among male college students, and PR statistically accounted for part of the cross-sectional PA–NSSI association. The two PA–PR profiles also differed in NSSI levels. These findings indicate that PA and PR are relevant correlates of NSSI, although their temporal and causal relationships require further investigation. Full article

Objective: To address the limitations of existing deep learning methods for brain networks in jointly modeling directional interregional connectivity, multiband information, and short-term dynamic features, this study proposes a multiscale directed temporal graph convolutional network (MSD-STGNN) for the three-class classification of healthy controls (HCs), patients with left temporal lobe epilepsy (lTLE), and patients with right temporal lobe epilepsy (rTLE). Methods: Resting-state magnetoencephalography (MEG) data were obtained from 43 subjects, including 14 HCs, 13 patients with lTLE, and 16 patients with rTLE. Based on 26 predefined default mode network (DMN)-related brain regions, directed effective connectivity networks were constructed in six frequency bands using the directed transfer function (DTF), and indices including information-flow strength, directional preference, and hemispheric asymmetry were used as node features. MSD-STGNN separately modeled incoming and outgoing connectivity information through directed graph convolution, fused frequency-band information using a hierarchical multiband attention mechanism with gated residual correction, and employed gated recurrent units (GRUs) to extract short-term dynamic features from consecutive brain-network slices. Model performance was evaluated using subject-level stratified fivefold cross-validation, with predictions from multiple temporal groups of each subject aggregated to obtain the final subject-level prediction. Frequency-band masking and node-level fusion-weight analyses were further performed to evaluate the model’s dependence on different frequency bands and information from the predefined brain regions. Results: In the subject-level fivefold cross-validation, MSD-STGNN achieved an accuracy of 0.836 ± 0.067, a macro-F1 of 0.830 ± 0.065, and a macro-AUC of 0.900 ± 0.055 using the one-vs-rest strategy, with the highest fivefold mean values across all evaluation metrics among the baseline models and ablation configurations investigated in this study. Post-training frequency-band masking showed that the model exhibited relatively high dependence on the low-gamma (30–80 Hz) and beta (13–30 Hz) bands. Node-level fusion-weight analysis showed that, within the 26 predefined DMN-related brain regions, orbitofrontal, cingulate, medial temporal, and parietal regions exhibited relatively high overall fusion weights across different frequency bands, although the exact top 5 regional rankings varied across folds. Significance: Within a unified framework, MSD-STGNN integrates directional connectivity, multiband information, and short-term dynamic features derived from MEG-based directed brain networks, providing a modeling approach with a certain degree of interpretability for the three-class classification of HCs, patients with lTLE, and patients with rTLE. The current findings are based on internal cross-validation of a small, single-center cohort; therefore, the classification performance and the observed frequency-band and brain-region attention patterns require further validation in larger, independent multicenter datasets. Full article

Retrofitting existing natural-gas power plants for hydrogen operation may reduce asset replacement, but energy-equivalent fuel substitution can substantially alter hydraulic loading and accident consequences. This study evaluates a 100% hydrogen fuel-supply train retrofit for an 8.11 MW gas-engine power plant by integrating hydraulic calculations, computational fluid dynamics (CFD), screening-level consequence analysis, hazard and operability analysis (HAZOP), and layer of protection analysis (LOPA). Under a hypothetical constant-energy-input scenario, the required normalized volumetric fuel flow increased from 1842.6 Nm³/h for natural gas to 6115.6 Nm³/h for hydrogen. In the existing 80A piping, the calculated velocity increased from 26.9 to 89.3 m/s; enlargement to 150A reduced hydrogen velocity to 22.8 m/s and the straight-pipe frictional pressure loss from 3.41 to 0.11 kPa per 10 m. CFD mean velocities were within approximately 3% of the analytical values and were interpreted as a numerical consistency check rather than independent validation. Screening-level ALOHA simulations produced slightly shorter hydrogen jet-fire radiation distances but substantially larger vapor-cloud-explosion (VCE) overpressure distances, including an 8.0 psi distance of 216 m under the documented scenario assumptions. HAZOP and LOPA identified residual scenarios and evaluated protection layers, with calculated mitigated frequencies below the project-specific 1.0 × 10⁻⁵ yr⁻¹ screening target under the adopted assumptions. These results indicate that a hydrogen fuel-supply train retrofit requires integrated evaluation of fuel demand, piping hydrodynamics, consequence characteristics, and risk-control measures rather than simple fuel substitution. Full article

Parkinson’s disease (PD) causes motor control deficiencies, resulting in gait irregularities such as shorter strides, slower walking speed, and irregular step timing. This study presents a video-based deep learning approach for PD classification that extracts skeletal keypoints from Timed Up and Go (TUG) test videos using AlphaPose and the COCO-17 representation. A total of 24 features were generated, comprising 23 conventional gait features and one Freezing of Gait (FoG) feature derived from frequency-domain analysis of ankle velocity signals. This FoG feature was not validated against clinician-confirmed FoG episodes and should be interpreted as a frequency-domain proxy rather than a diagnostic measure. Three feature selection procedures and four LSTM-based architectures were evaluated across full, walking, and turning segments. Experimental results on a self-collected TUG dataset showed that the Standalone FI configuration achieved the numerically highest test accuracy of 77.78% on the turning segment among the evaluated LSTM configurations, while conventional features achieved 66.67% on both the full and walking segments. These test-set metrics provide descriptive estimates derived from a static subject-level test division involving six held-out participants (excluded from training and validation) and should not be viewed as statistically dependable indicators of clinical performance at the population level; in addition, gait-cycle boundaries were not independently validated and fallback usage was not quantified. Turning segments demonstrated higher discriminative power than straight-walking segments. Zero-shot cross-dataset evaluation on Turn-REMAP and PD-Walk revealed a substantial generalization gap, with accuracy falling to 55.12% and 49.53%, respectively, indicating that the present model is not yet suitable for cross-site clinical deployment without adaptation or calibration. The proposed framework provides systematic insights into the comparative role of conventional and FoG-derived gait parameters for non-invasive video-based PD screening. Full article