Background: Three-dimensional gait analysis, supported by advanced sensor systems, is a crucial component in the rehabilitation assessment of post-stroke hemiplegic patients. However, the sensor data generated from such analyses are often complex and challenging to interpret in clinical practice, requiring significant time and complicated procedures. The Gait Deviation Index (GDI) serves as a simplified metric for quantifying the severity of pathological gait. Although isokinetic dynamometry, utilizing sophisticated sensors, is widely employed in muscle function assessment and rehabilitation, its application in gait analysis remains underexplored. Objective: This study aims to investigate the use of sensor-acquired isokinetic muscle strength data, combined with machine learning techniques, to predict the GDI in hemiplegic patients. This study utilizes data captured from sensors embedded in the Biodex dynamometry system and the Vicon 3D motion capture system, highlighting the integration of sensor technology in clinical gait analysis. Methods: This study was a cross-sectional, observational study that included a cohort of 150 post-stroke hemiplegic patients. The sensor data included measurements such as peak torque, peak torque/body weight, maximum work of repeated actions, coefficient of variation, average power, total work, acceleration time, deceleration time, range of motion, and average peak torque for both flexor and extensor muscles on the affected side at three angular velocities (60°/s, 90°/s, and 120°/s) using the Biodex System 4 Pro. The GDI was calculated using data from a Vicon 3D motion capture system. This study employed four machine learning models—Lasso Regression, Random Forest (RF), Support Vector regression (SVR), and BP Neural Network—to model and validate the sensor data. Model performance was evaluated using mean squared error (MSE), the coefficient of determination (R²), and mean absolute error (MAE). SHapley Additive exPlanations (SHAP) analysis was used to enhance model interpretability. Results: The RF model outperformed others in predicting GDI, with an MSE of 16.18, an R² of 0.89, and an MAE of 2.99. In contrast, the Lasso Regression model yielded an MSE of 22.29, an R² of 0.85, and an MAE of 3.71. The SVR model had an MSE of 31.58, an R² of 0.82, and an MAE of 7.68, while the BP Neural Network model exhibited the poorest performance with an MSE of 50.38, an R² of 0.79, and an MAE of 9.59. SHAP analysis identified the maximum work of repeated actions of the extensor muscles at 60°/s and 120°/s as the most critical sensor-derived features for predicting GDI, underscoring the importance of muscle strength metrics at varying speeds in rehabilitation assessments. Conclusions: This study highlights the potential of integrating advanced sensor technology with machine learning techniques in the analysis of complex clinical data. The developed GDI prediction model, based on sensor-acquired isokinetic dynamometry data, offers a novel, streamlined, and effective tool for assessing rehabilitation progress in post-stroke hemiplegic patients, with promising implications for broader clinical application.

Objective: This study aimed to investigate the muscle activation of patients with lumbar disc herniation (LDH) during walking by surface electromyography (SEMG) and establish a diagnostic model based on SEMG parameters using random forest (RF) algorithm for localization diagnosis of compressed nerve root in LDH patients.Methods: Fifty-eight patients with LDH and thirty healthy subjects were recruited. The SEMG of tibialis anterior (TA) and lateral gastrocnemius (LG) were collected bilaterally during walking. The peak root mean square (RMS-peak), RMS-peak time, mean power frequency (MPF), and median frequency (MF) were analyzed. A diagnostic model based on SEMG parameters using RF algorithm was established to locate compressed nerve root, and repeated reservation experiments were conducted for verification. The study evaluated the diagnostic efficiency of the model using accuracy, precision, recall rate, F1-score, Kappa value, and area under the receiver operating characteristic (ROC) curve.Results: The results showed that delayed activation of TA and decreased activation of LG were observed in the L5 group, while decreased activation of LG and earlier activation of LG were observed in the S1 group. The RF model based on eight SEMG parameters showed an average accuracy of 84%, with an area under the ROC curve of 0.93. The RMS peak time of TA was identified as the most important SEMG parameter.Conclusion: These findings suggest that the RF model can assist in the localization diagnosis of compressed nerve roots in LDH patients, and the SEMG parameters can provide further references for optimizing the diagnosis model in the future.

Background: Patients with Parkinson’s disease (PD) commonly experience difficulties when performing a second task while walking. The mechanisms underlying dual-task walking deficits remain poorly understood. In previous studies the second tasks were often simplistic, typically comprising questions from standardized cognitive assessments. Additionally, existing fNIRS studies comparing PD patients and healthy controls have reported inconsistent findings, limiting our understanding of prefrontal cortex (PFC) contributions to cognitive-motor integration.Methods: Forty-two healthy older adults (15 men and 27 women, age 59.97 ± 5.58 years) and fifty-eight patients with PD (25 men and 33 women, age 61.07 ± 7.56 years, Hoehn and Yahr stage 1∼3) were enrolled. The protocol consisted of three repetitions of these conditions: stationary marching and marching while two-digit arithmetic calculating. Researchers used fNIRS to measure PFC activation and changes in △HbO2 concentration during tasks execution.Results: Healthy controls demonstrated task-dependent prefrontal modulation - selective activation (6/22 channels, p < 0.05) during single-task conditions contrasted with global prefrontal engagement (22/22 channels, p < 0.05) under dual-task demands. In contrast, PD patients showed widespread activation across all 22 channel regions during both single and dual tasks (p < 0.05). During task switching, healthy subjects experienced significant activation increases in 15/22 channel regions (p < 0.05), paralleled by significant rises in ΔHbO2 concentrations across five prefrontal regions (p < 0.05; Cohen’s d ranging from 0.43 to 0.82). Conversely, PD patients exhibited no significant difference in the activation of all 22 channel regions (p > 0.05), and no significant changes of ΔHbO2 concentrations across all regions between single and dual tasks (p > 0.05; Cohen’s d < 0.30).Conclusion: Findings indicate that simple marching tasks underengage prefrontal resources in healthy individuals, whereas dual tasks engage greater prefrontal activation to meet heightened cognitive demands. In contrast, owing to disruptions in the cortico-basal ganglia-thalamocortical circuitry, PD patients exhibit a “ceiling effect” in PFC activation: increased task difficulty fails to elicit proportional activation, likely because single tasks already overtax prefrontal resources. This divergence in neural adaptability underscores core differences in cognitive-motor integration mechanisms between healthy individuals and PD patients, providing a basis for developing targeted dual-task interventions to enhance neural efficiency.

Population ageing and chronic disease are increasing demand for rehabilitation, while resources remain limited. This review does not report an implemented end-to-end system; instead, it proposes a modular workflow framework for applying large AI foundation models across rehabilitation. Organised into four stages—assessment, prescription, execution, and monitoring—we summarise recent evidence and highlight techniques most suitable at each stage. In assessment, multimodal models can enable more continuous and objective functional measurement from heterogeneous sensor and imaging data. In prescription, large language models can support evidence-informed, personalised plan formulation by synthesising guidelines and patient context. In execution, vision–language–sensor models can provide real-time feedback for telerehabilitation and adherence support. In monitoring, longitudinal and cross-setting data integration can facilitate risk prediction and early warning for safety and long-term management. We also discuss practical adaptation options (e.g., parameter-efficient fine-tuning) and propose a clinimetric-oriented evaluation framework to assess validity, reliability, and generalisability. By mapping AI capabilities to concrete workflow tasks, the framework provides a theoretical foundation and roadmap for reproducible research and future translation toward a universal rehabilitation model.

To the Editor In a recent article, Stickel et al contributed to exploring the association of hearing loss with 7-year cognitive change and mild cognitive