高齢者のロコモティブシンドローム評価指標と実動作計測による身体機能評価の比較

抄録

In this paper, we propose a machine learning method based on motion capture system. The method extracts features from various kinds of data such as force plate data and motion capture data. We filtered the extracted features so that we can compare each data on the same number line. Comparing this method and clinical evaluation method, it is cleared that we could align 3 groups (non-frailty, pre-frailty, and frailty) on the same number line. The possibility of grouping into more groups than clinical evaluation method for frailty is shown by plotting within-class average of each participants’ data. Applying our filter to various motion data, we can compare frailty level more personally and more numerically.