Clustering algorithm-based prediction model for athlete performance grading and injury risks

Abstract

For athlete performance evaluation and injury risk prediction—which is increasingly crucial—traditional approaches find difficulty handling complex, multidimensional data. We introduce the PerfoRisk-KDB model to precisely estimate athlete performance and injury risk by combining K-means and DBSCAN clustering techniques. By combining these two clustering techniques, the idea of this work surpasses the constraints of a single technique and increases accuracy and robustness for complex and high-dimensional data. This work tests the performance assessment and injury risk prediction of a real athlete dataset against conventional models. Based on tests, the PerfoRisk-KDB model shows good performance on several evaluation criteria and shows good application possibilities.