Webcam-Based Exergame for Motor Recovery with Physical Assessment via DTW.

The single-angle feature extraction method provided the clearest distinction between normal and abnormal movements among the three methods evaluated.

• Three feature-extraction methods were compared in the pilot study
• The single-angle method was particularly effective for abduction and elbow flexion movements
• Score separation for shoulder flexion was less distinct using this method
• Movements were simulated using a 5-kg resistance band to approximate abnormal movement conditions

Shoulder flexion assessment showed less distinct score separation between normal and abnormal movements compared to abduction and elbow flexion.

• Reduced score separation for shoulder flexion was attributed to movement variability and posture-related angle fluctuations
• The authors indicate this suggests further refinement of feature design is needed
• This was observed despite applying DTW together with a Z-score-based scoring model
• FMA-UE (Fugl-Meyer Assessment Upper Extremity) guidelines were used as the basis for movement quantification

RehabHub uses Dynamic Time Warping (DTW) combined with a Z-score-based scoring model to quantify upper-limb movement quality.

• The system specifically assesses abduction, shoulder flexion, and elbow flexion
• Assessment is based on FMA-UE (Fugl-Meyer Assessment Upper Extremity) guidelines
• The Z-score-based scoring model relies on data from non-clinical adult participants
• MediaPipe is used for real-time pose estimation via a common webcam

The RehabHub system integrates a cloud-based platform that supports remote monitoring for caregivers.

• The platform gives caregivers access to both performance scores and recorded exercise videos
• The system is designed for home-based use requiring only a common webcam
• The approach is characterized as low-cost
• The system supports exergaming combined with standardized physical assessment directly in gameplay

A pilot study evaluated the feasibility of webcam-based movement assessment by simulating abnormal movements with a resistance band.

• A 5-kg resistance band was used to simulate movements approximating abnormal movement conditions
• Three feature-extraction methods were evaluated in the pilot study
• The study included non-clinical adult participants for reference data collection
• The results demonstrate 'the feasibility of a low-cost webcam-based assessment integrated into exergaming'