OPTIMIZATION OF ATHLETE INJURY PREVENTION STRATEGIES BASED ON BIG DATA ANALYSIS

Abstract

Injuries sustained by athletes during sports can have a significant impact on their career and physical health. To reduce the damage caused by athletes' injuries, this study combines three-dimensional convolutional neural networks and attention mechanisms to extract key features and identify normal and abnormal behaviors of marathon athletes. By improving the residual network structure and introducing attention mechanisms, the model can more effectively focus on key features in videos. In the research results, the accuracy curve of the 16 sampled frames used converged at 30 iterations, and the accuracy converged to around 0.884, which was significantly higher than the other network model groups. In addition, the proposed model performed well in identifying normal and abnormal behaviors of marathon athletes, with an accuracy rate of up to 94.69%. Especially for the four types of marathon athletes, the recognition accuracy was above 90%. The recognition accuracy of belly covering behavior was the highest, reaching 93.29%. The experiment showed that the constructed network model could timely and accurately identify the normal and abnormal movement behaviors of athletes. This study can provide scientific basis for athlete injury prevention and performance improvement, which can help improve the overall competitive level and health status of athletes.