Strokes are the leading cause of disability in France. Motor rehabilitation is crucial to reducing the effects of strokes. Although rehabilitation with a physiotherapist is essential, their availability is limited. This is why telerehabilitation is an important complement to post-stroke care. The challenges of telerehabilitation include the difficulty for patients to verify that exercises are being performed correctly and to maintain the motivation to follow their rehabilitation programme regularly.

This thesis is a collaboration between the Loki research team, specialising in Human-Machine Interaction (HMI), and Myodev, a company developing a rehabilitation solution based on EMG sensors. Our goal is to improve EMG telerehabilitation by accurately measuring muscle activity to ensure that exercises are performed correctly. In addition, we want to integrate gamification elements to encourage patients to follow their rehabilitation programme regularly.

The first phase aims to improve EMG signal processing, building on the candidate's master's degree internship. Initially, we will automatically and accurately detect the characteristics of a muscle contraction, such as its start, end and intensity. Next, we will optimise the algorithm developed during the internship to ensure effective real-time detection of muscle contractions. To do this, we will collect a set of data to obtain a reference, adjust the algorithm parameters and validate the automatic detections.

In the second phase, we will develop a telerehabilitation application incorporating gamification elements. We will base this on exercises validated by physiotherapists that patients can perform independently, with automatic validation of progress thanks to our muscle contraction detection and characterisation system. We will design gamification elements to encourage patients to follow their rehabilitation programme. One initial idea is to create forms of real-time biofeedback. Another idea would be to use the characteristics of contractions as input events for video games. The effectiveness of the new methods will be validated by comparative studies with users.

A third phase will focus on integrating additional sensors (IMU or other) to better track patient movements. This will involve combining information from the various sensors to estimate patient movements and adapt feedback and gamification.

All this work will be validated with therapists and patients through laboratory and field experiments, which will be subject to validation by an ethics committee.

1. State of the art in telerehabilitation, particularly through gamification
2. Improvement of the algorithm for detecting muscle contractions on an EMG signal
3. Experimental validation of the detection algorithm and its parameters
4. Biofeedback design
5. Design of video games controlled by muscle contractions
6. Comparative study of gamification methods