Harbin University

ABSTRACT Neural networks have the potential in approximating the solutions of the Navier–Stokes equations. To address the high computational complexity of high‐order derivatives based on automatic differentiation and the high memory requirements of deep neural network training, a novel Deep Collocation Method (DCM) that combines reversible neural network (RevNet) and spectral differentiation is proposed. The memory‐efficient RevNet is employed to approximate the solutions, while Chebyshev differentiation matrices are derived to rapidly and accurately compute the spatiotemporal derivatives of the approximate solutions, making the computational complexity of high‐order derivatives comparable to that of the forward pass, and significantly reducing memory requirements. A unified spectral discretization scheme is applied in both temporal and spatial dimensions, enabling seamless integration of the neural network and the spatiotemporal Chebyshev differentiation matrices. It effectively incorporates global spatiotemporal information into the residual computation of the governing equations at each collocation point, enabling the neural network to actively account for nonlocal effects of the flow field. Validation results demonstrate that DCM performs well various benchmark flow problems and enables efficient training of deep models even on memory‐constrained devices.

BackgroundCombat sports involve repeated high-intensity, intermittent actions under direct competition. During competition, athletes must produce high power output repeatedly, with rapid energy-system transitions and substantial oxidative stress responses.ObjectiveThis cross-sectional study examined associations between antioxidant-related indicators and special strength performance in combat athletes.MethodsA cross-sectional study design was used, recruiting 25 high-level combat athletes in training. We measured one energy-related marker [plasma adenosine triphosphate (ATP)], three antioxidant defense markers [total antioxidant capacity [T-AOC], superoxide dismutase [SOD], glutathione peroxidase [GPX]], and one oxidative damage marker [malondialdehyde (MDA)] and six special strength indicators [standing long jump, vertical jump, medicine ball throw, one-repetition maximum (1-RM) bench press, one-repetition maximum (1-RM) squat, 30 m sprint] were measured. Gender, height, weight, body fat percentage, and resting heart rate were included as covariates. Shapiro Wilk tests, Kruskal Wallis and Mann Whitney U tests, Spearman correlations, multiple linear regression, and Bayesian regression were applied.ResultsBody fat percentage and SOD activity differed across sports (p < 0.05), with higher body fat percentage in taekwondo and higher SOD activity in boxing. In gender comparisons, females had lower T-AOC than males (p = 0.011), with no significant differences for the other indicators. ATP was positively correlated with standing long jump (r = 0.449) and vertical jump (r = 0.502), while body fat percentage showed a negative correlation with vertical jump (r = −0.529, p < 0.05). None of the multiple linear regression models reached statistical significance. Bayesian regression yielded similar patterns, with positive posterior means for ATP in explosive and strength outcomes and negative estimates for MDA in some models.ConclusionSpecial strength outcomes were associated with markers related to energy availability and oxidative status. ATP and body fat percentage were the variables most consistently associated with explosive performance in this sample: ATP was positively associated with standing long jump and vertical jump, whereas body fat percentage was negatively associated with vertical jump performance. Differences in antioxidant indicators across sports and between sexes suggest that antioxidant status may vary with individual characteristics and sport-specific demands. These findings are hypothesis-generating and may inform future work on training, nutritional strategies, and pre-competition preparation, which should be tested in longitudinal or randomized controlled studies.

BackgroundParkinson disease (PD) impairs gait, balance, and quality of life, and wearable devices have been proposed to support rehabilitation, but evidence for their clinical efficacy remains uncertain.ObjectiveThis study aimed to evaluate, within the International Classification of Functioning, Disability, and Health (ICF) framework, the effects of wearable-device interventions on gait performance, balance, and health-related quality of life in people with PD by conducting a systematic review and meta-analysis of randomized controlled trials (RCTs).MethodsWe searched PubMed, Web of Science, Cochrane Library, Embase, and ClinicalTrials.gov from inception to November 18, 2025, for RCTs in people with PD comparing wearable-device interventions with control conditions. We used Hartung-Knapp random-effects models to pool mean differences (MDs) or standardized mean differences (SMDs) and reported 95% prediction intervals when ≥3 studies were pooled. Risk of bias was assessed using the Cochrane Risk of Bias (RoB) tool, and certainty of evidence was rated using Grading of Recommendations Assessment, Development, and Evaluation (GRADE).ResultsNine RCTs involving 260 participants were included. Wearable devices produced a small improvement in stride length (MD 0.10 meter, 95% CI 0.03-0.17), but there was no clear benefit for the 10-Meter Walk Test time (MD 0.04 second, 95% CI –0.06 to 0.15). Double support time showed no reduction (MD –1.59% gait cycle, 95% CI –3.79 to 0.61). Freezing of gait (Freezing of Gait Questionnaire [FOG-Q] and New Freezing of Gait Questionnaire [NFOG-Q]) did not significantly improve (SMD –0.24, 95% CI –0.72 to 0.24). Motor severity (Unified Parkinson Disease Rating Scale Part III [UPDRS III]) showed a small, nonsignificant trend favoring wearable devices (MD –2.16 points, 95% CI –4.39 to 0.07). For balance, pooled results from the Berg Balance Scale (BBS), Mini Balance Evaluation Systems Test (Mini-BESTest), and Performance-Oriented Mobility Assessment Balance Subscale (POMA balance) suggested a borderline effect (SMD 0.48, 95% CI –0.02 to 0.98). Wearable devices did not meaningfully improve Parkinson Disease Questionnaire (PDQ) scores (SMD –0.28, 95% CI –0.74 to 0.17), EQ-5D utility (MD 0.10, 95% CI –0.24 to 0.44), or Falls Efficacy Scale–International (FES-I) scores (MD –0.04, 95% CI –1.10 to 1.02). Prediction intervals frequently crossed the null, suggesting effects may vary by setting and population.ConclusionsWearable device interventions for Parkinson disease produced a small improvement on average in stride length, with no consistent benefits for other gait outcomes, balance, or patient-centered outcomes. By integrating ICF mapping with Hartung-Knapp meta-analysis, prediction intervals, and GRADE, and avoiding pooling of conceptually distinct gait measures used in prior reviews, this review clarifies where evidence is most consistent, supports using wearables as adjuncts to rehabilitation, and underscores the need for larger, longer RCTs with standardized outcomes to determine who benefits and how to implement them.Trial RegistrationPROSPERO CRD42024585686; https://www.crd.york.ac.uk/PROSPERO/view/CRD42024585686