Spatio-temporal situation recognition for groups in caregiving services

Recently, various kinds of ICT devices are introduced into caregiving services. One of the benefits of using ICT devices is that use logs including time, location, and various data on staff's activities are automatically collected by the devices, and valuable information for improving the work can be extracted from the logs. In this paper, we propose a method for automatically recognizing operation situations from the use logs based on the data science techniques. What's new in the proposed approach is the two-phase clustering that reflects both spatio and temporal information. In the first phase, vectors that represent the arrangement of staff are clustered in order to recognize spatial situation. In the second phase, change of the staff's arrangement on the timeline is taken into consideration. The proposed method is applied to use logs collected during the field experiments which was conducted in a nursing home. The result is examined by visualizing the time series of situations, and comparing them with the actual situations.