Automatic monitoring of rehabilitation sessions through inertial measurements

The analysis of human movement is crucial in medical and biomechanical research, particularly for assessing motor impairments and rehabilitation strategies. In recent years, the use of sensors in physiotherapy has enabled objective evaluation of patient performance. Inertial measurement units (IMUs) offer a portable and reliable solution for movement analysis, allowing the extraction of joint angles, acceleration, and velocity when placed in key body segments. This thesis presents the design and implementation of algorithms to isolate and classify exercises performed during physiotherapy sessions. A dataset of healthy subjects and patients was collected using IMUs placed on key body segments. Data processing techniques, including autocorrelation analysis and decision trees, were applied for robust activity classification. Additionally, lateral and backward gait were analyzed to automatically segment the gait cycle, testing machine learning and rule-based approaches. Preliminary results indicate that IMUs can effectively characterize rehabilitation sessions and provide unbiased insights into exercise performance. These findings have potential applications in clinical assessments, rehabilitation monitoring, and movement disorder analysis, offering valuable support to healthcare professionals and researchers.