Figure 1.

The model generation process proceeds sequentially from (a) to (d). (a) refers to collected biometric sensor data and questionnaire data. In (b), it refers to the pre-processing of the data collected in (a). The 3-axis accelerometer sensor data, BVP data, and questionnaire data, which are raw data of each sensor, are applicable. In (c), two types of features are extracted from data that has undergone pre-processing. These include features related to depression symptoms and statistical features. The general model and the personal model are generated from features and modified labels as shown in (d).