A Skin-Adherent Magneto-Inertial Wearable for Real-Time Joint Motion Analysis

Abstract

Musculoskeletal physiotherapy is evolving with the integration of wearable technologies that enable continuous monitoring and personalized rehabilitation. This study presents a multimodal, wireless, and cost-effective wearable system designed to assess and classify joint motion using magneto-inertial sensing. The system incorporates a flexible PDMS-NdFeB magnetic skin patch and a compact sensor module with a 9-degree-of-freedom IMU. Real-time joint kinematics are wirelessly transmitted to a custom mobile application, enabling interactive visualization and feedback. Biomechanical modeling is employed to evaluate muscle contributions and joint dynamics across the wrist, elbow, and knee. Processed magnetic signals are used to classify range of motion (ROM) into three categories—Limited, Normal, and Hypermobility—through traditional machine learning and deep learning models. A 1D convolutional neural network (1D-CNN) achieves the highest classification accuracy (95.3%). The proposed system demonstrates strong potential for enhancing musculoskeletal rehabilitation by providing accurate, real-time assessments and individualized treatment feedback.