머신러닝 기반 맨발 및 운동화 착용 시 하지 근육 활성도 비교 분석을 통한 보행 상태 분류 연구

Abstract

This study aims to compare the muscle activity of the lower limbs between barefoot and shod walking and to classify the two walking conditions using a machine learning approach. Walking is a fundamental human activity, and muscle activation patterns can vary depending on external conditions such as footwear. Despite the abundance of studies comparing different types of shoes, there is a lack of research focusing on quantitative differences between barefoot and shod walking. In this study, surface electromyography (sEMG) signals were collected from four lower limb muscles (Tib- ialis Anterior, Rectus Femoris, Vastus Medialis, and Medial Gastrocnemius) from five healthy young adults while they walked under both barefoot and shod conditions. The sEMG signals were preprocessed, segmented using gait event markers, and then analyzed to extract time-domain features including Root Mean Square (RMS), Mean Absolute Value (MAV), Waveform Length (WL), and Variance (VAR). A Random Forest classifier was then trained to distinguish between the two walking conditions. The results showed distinct differences in muscle activation patterns between the two conditions, and the classification model achieved an accuracy of approximately 95%. These findings suggest that machine learning models based on sEMG features can effectively differentiate gait conditions and may contribute to future personalized gait assessment and footwear evaluation systems.