Personalized real-time sleep stage remote monitoring system

This paper proposes the real-time remote monitoring system for personalized sleep stage. Such a system enables care workers to monitor a sleep stage of aged persons remotely, which is required to support aged persons who tend to wander in midnight and/or who are easily to fall down from their beds. For this purpose, we develop the system that can estimate a sleep stage in real-time from a heartbeat and body movement data acquired by an unconstraint-typed pressure sensor, and such a sleep stage is remotely displayed in tablets of care worker to inform a care timing for aged persons. Intensive human subject experiments have revealed the following implications: (1) the past data for estimating the sleep stage (even in one day data) contributes to estimating the future sleep stage; (2) the precision of the sleep stage estimation increases as the number of past data increases; and (3) such a precision further increases by using either of the past data where a heartbeat rate gradually increases or decreases. (i.e., the past data where a heartbeat rate increases (or decreases) contributes to improving a precision of the sleep stage of a person whose heartbeat rate increases (or decreases)).