Sensory Interference Gait Classification Based on HLOA-HFWA

Human gait balance relies on the coordination of three sensory systems: vision, proprioception, and the vestibular system. Disruptions in these systems can lead to gait abnormalities. Existing detection methods primarily focus on single-sensory interference, lacking research on distinguishing multiple types of sensory disturbances. This study proposes a non-invasive, interference-free gait detection framework based on gait signals collected using Qualisys. Time-domain, frequency-domain, wavelet, and nonlinear features were extracted; however, feature redundancy increases the complexity of learning and classification. To address this, we introduce a filter-wrapper hybrid feature selection algorithm (HLOA-HFWA) based on HLOA-optimized feature aggregation distance, aiming to efficiently select the optimal feature subset and enhance classification performance. Experimental results demonstrate that the proposed method outperforms existing approaches in classification accuracy, verifying its effectiveness and robustness.