Chronic Stress Level Estimation Focused on Motion Pattern Changes Acquired from Seat Pressure Distribution

In modern society, chronic stress in the workplace is a serious problem because it causes numerous diseases. To prevent chronic stress, it is necessary to continuously measurement this type of stress in the workplace. Thus, it is necessary to realize a chronic stress estimation method with small load to users. Therefore, we proposed a method to estimate chronic stress level by measuring changes of body motion patterns during desk work caused by chronic stress. For the acquisition of body motion patterns we called "Motion Pattern", a cushion-type sensor device with six pressure sensors was developed, and the changes of seat pressure distribution during desk work were measured. "Motion Pattern" was the time-series pattern of four"Motion Labels" including "Impact Label" which represented the motion that impacts the seat. We tried to estimate chronic stress level by learning the relationship between this "Motion Pattern" and the score of stress evaluation questionnaire. As a result, it was found that there was a significant difference in the interval time of "Impact Label" occurrence in all six participants. In addition, we achieved to estimate the existence and non-existence of chronic stress with an average accuracy of 87.0% using a classifier learned for each individual and achieved to estimate chronic stress level with a maximum accuracy of 70.0%for one participant. It suggests that our proposed method has potential for chronic stress monitoring in the workplace.