Feature Selection Methods in Walking Stability Analysis

Walking stability is the main reason for leading to falls for people, especially for elders. But there are much more features related with walking so that we cannot understand which features are more important than others to contribute the walking stability. Almost all of researches focused on some specific features but didn´t present any reasons for that. Therefore, the Dynamic Time Warping (DTW) is employed to calculate walking stability, an adaptive Genetic Algorithm (GA) to search the best contributing and representative features, and an improved Support Vector Machine (SVM) to assess the fitness of specific feature combination according to age classification information. After studying walking patterns of 51 healthy male subjects ranging from 21 years old to 66 years old, the 32 most contributing features are acquired. The experiments show that these feature selection methods can improve the discrimination power of walking stability effectively. This research not only supports more convenient data acquisition equipments, but also helps us understand walking stability better.