Estimation of vital signs in bed from a single unobtrusive mechanical sensor: algorithms and real-life evaluation.

A single contact-less mechanical sensor is exploited for estimating three vital signs during sleep, namely, the heart rate, the breathing rate and an activity index related to the body movements. Robust estimations are achieved over epochs of 30 seconds. The data processing is performed with standard DSP techniques leading to an integrated solution for dealing with body motion artifacts. The algorithms are described and evaluated over a one-hundred night corpus collected from real-life recordings of healthy subjects and sleep-laboratory patients. Results show that the average error of the heart rate is of 1.25 beat per minute compared to the reference values from an ECG and the coverage of the mechanically derived estimates is 83%.