Application of sorted metrics based on insole feature mining

Abstract

To improve the accuracy and real time performance of fall risk assessment, this study proposes a feature selection method based on sorted metrics. The method evaluates previously generated spatiotemporal features using correlation coefficient (CCF), bayes factor (BF), and selfinformation index (SII) to identify the most informative features. The results show that these metrics play complementary roles in the feature selection process. CCF measures the linear correlation between features and helps eliminate redundancy. BF emphasizes the statistical significance between high-risk and low-risk groups. SII captures distributional differences from the perspective of information entropy. Comparative analysis across various metrics demonstrates that the combined method achieves high accuracy across multiple performance indicators while significantly reducing computational complexity.