Sports fatigue detection based on deep learning

Moderate exercise is good for human health. However, when the exercise intensity exceeds a certain level, it will be harmful to the human body. Therefore, precise control and adjustment of exercise load can ensure athletes' sports safety and improve their competitive performance. In this work, we have developed wearable exercise fatigue detection technology to estimate the human body's exercise fatigue state using real-time monitoring of the ECG signal and Inertial sensor signal of the human body. 14 young healthy volunteers participated in the running experiment, wearing ECG acquisition equipment and inertial sensors. ECG, acceleration and angular velocity signals were collected to extract features. And then Bidirectional long and short-term memory neural network (Bi-LSTM) was used to classify three levels of sports fatigue. The results showed that the recognition accuracy of the user-independent model was 80.55%. The experimental results verified the effectiveness of the algorithm.