Automatic Gait Classification using Arm Acceleration based on Stature and Gait Speed

Unsuitable foot clearance and step length lead falling accidents; thus, these gait parameters should be monitored for fall prevention. According to previous study, users prefer wrist as a position to wear wearable device for daily gait monitoring. Therefore, we have been developing gait monitoring method using wearable device on a wrist. Our previous method classifies gaits with different foot clearance or step length by machine learning technique using 3-axis acceleration data obtained from wearable device on a wrist. The results of our previous study showed that this method could classify gaits for each step. However, accuracy of our previous method was insufficient because individual difference of arm acceleration due to stature or gait speed were not considered. The objective of this study is to propose the gait classification method based on individual difference. The proposed method classifies three gaits with different foot clearance and step length by machine learning technique using combination of arm acceleration, stature, and gait speed. The proposed method was tested for 1800 steps performed from ten participants. The proposed method could classify three gaits with accuracy greater than 0.9. These results indicate possibility that the proposed method can be used for gait monitoring using wearable device.