高齢者の大規模調査データからの虚弱指標の探索

抄録

In this paper, we proposed the coordinate conversion method using matrix decomposition. Using this method, we can associate indices that is find out from the small-scale, precise dataset with the large-scale dataset. Vice versa, we can place each set of data in the small-scale dataset into the large-scale dataset, and can interpret as statistically. Kashiwa study dataset that is a cohort dataset for frailty is selected as the large-scale dataset. Finding out the physical/cognitive frailty axis from our own small-scale dataset of features extracted from several measurements, the transformation matrix was calculated from features of both the large-scale and the small-scale dataset. Though there are several limitations, we could extract features that are related only with the cognitive frailty from the large-scale dataset and we were able to compare data from two different datasets in the same feature space.