An object of the present invention is to accurately estimate an index value representing a state of the mind and body of a user without directly measuring and even when there is an individual difference in biometric data between users. In a learning phase, activity state data and biometric data at that time are acquired from user terminals of a plurality of users, respectively, and based on these measurement data, biometric data and activity state are obtained by using a regression analysis technique. A regression equation representing the relationship with the data is generated, the difference between the coefficients of the regression equation between all users and each user is obtained, and the coefficient correction regression representing the relationship between the difference between the coefficients and the average value of the biometric data. Generate an expression. In the estimation phase, the biometric data of the new user is acquired, and the coefficient value of the regression equation for estimating the activity state is calculated based on the average value of the biometric data and the regression equation for coefficient correction. And the activity state of the new user is estimated using the regression equation after the coefficient correction. [Selection diagram] Fig. 1【課題】ユーザの心身に係る状態を表す指標値を、直接計測することなく、かつユーザ間で生体データに個体差がある場合でも、精度良く推定する。【解決手段】学習フェーズにおいて、複数のユーザのユーザ端末から活動状態データとそのときの生体データをそれぞれ取得し、これらの計測データをもとに、回帰分析の手法を用いて生体データと活動状態データとの関係を表す回帰式を生成し、当該回帰式の係数について全ユーザと各ユーザとの間の差分を求め、この係数の差分と生体データの平均値との関係を表す係数補正用回帰式を生成する。そして、推定フェーズにおいて、新規ユーザの生体データを取得し、当該生体データの平均値と上記係数補正用回帰式をもとに活動状態推定用の回帰式の係数値を上記新規ユーザ用の係数値に補正し、この係数補正後の回帰式を用いて新規ユーザの活動状態を推定する。【選択図】図１
