Self-organizing maps in adaptive health monitoring

A method for health monitoring is considered. Measured physical signals have been dynamically classified to low-, middle- or high-levels and a self-organizing map (SOM) has been utilized to combine the information. The data were collected during spring 1996 and consist of over eight weeks of physical measurements and diaries recorded in a home environment by four test subjects. The research shows that this method can be used to monitor the system of a human being. The system finds some daily structures as well as differences between weekdays and weekend. The physical activities have much stronger effect on the signals than mental stress states, which show no clear clustering on maps.