Evaluation Model of Running Fatigue of Young Students Based on Characteristic Parameters of ECG Signal

Large or high-intensity running training can lead to fatigue and sports injuries; therefore, research on sports fatigue and the establishment of a fatigue analysis and evaluation model for running sports are of guiding significance for the design and training of running sports tasks. 22 subjects participated in accelerated running experiments from the slowest 8 km/h to the fastest 13 km/h, and the RPE (rating of perceived exertion) and ECG signal data were recorded. After 3 days, all subjects ran at a constant speed between 13 and 15 RPE during the acceleration run and the RPE and ECG signals were measured. To address the problem of motion artifacts in the ECG signal during exercise, a 7-layer “db8” smooth wavelet transform was used to extract ECG features. Then, five HRV (heart rate variability) indicators, NN-mean, SDNN, LF, HF, and TP, which are more sensitive to fatigue changes, were selected by applying single factor analysis of variance and single feature linear regression methods. The results showed that all five HRV indexes decelerated and decreased with the increase in exercise time, and finally reached a stable state, i.e. fatigue. Based on the subjective evaluation data and the five HRV characteristic indexes obtained, a tri-classification model for fatigue prediction was established by using a support vector machine (SVM), and its fatigue prediction accuracy was verified to be 91.06%, which can effectively evaluate running fatigue.