Congenital Talipes Equinovarus (CTEV), or clubfoot, affects approximately 1–2 per 1,000 live births worldwide. Although the Ponseti method achieves correction rates exceeding 90%, long-term outcomes depend on sustained adherence to physiotherapy and bracing, which is often suboptimal and difficult to monitor outside clinical settings. This study presents ToeFro-Varus, a proof-of-concept system that integrates an ankle-worn inertial measurement unit (IMU) with a gamified mobile application to enable automated monitoring of prescribed physiotherapy exercises in school-age children (5–18 years) in the maintenance phase of treatment. The wearable device, built using an ESP32 microcontroller and a BNO055 9-axis IMU, captures motion data across six clinically relevant exercises. A supervised machine learning pipeline using engineered time-series features was evaluated with a participant-wise split (17 participants: 11 training, 6 testing). The best-performing model (Gradient Boosting) achieved 95.9% classification accuracy on held-out participants; however, evaluation relied on a limited number of original trials (n = 36), with additional augmented samples used for training, and therefore these results should not be interpreted as evidence of generalization.This work demonstrates the technical feasibility of IMU-based exercise classification coupled with real-time gamified feedback for remote monitoring. The system enables objective identification of exercise type and provides a framework for linking physical activity to digital rewards. However, the study does not assess adherence, quality of execution, or clinical outcomes, and the generalizability of the model remains unestablished. The primary contribution is an end-to-end, low-cost prototype illustrating a scalable approach to automated physiotherapy monitoring, providing a foundation for future studies evaluating clinical effectiveness and real-world deployment.
Nadkarni et al. (Sat,) studied this question.