Toward Sustained Logging in Sleep Studies

through Energy-Efficient Actigraphy

by Marko Borazio

Abstract

This thesis presents a wearable sensing platform and set of preliminary studies which evaluate the use of data from simple, energy-efficient sensors for long-term monitoring of sleep. The approach is based on actigraphy, which describes the possibility to record a person’s movements over a period of time. It is often used as a support in the medical environment when patients are observed, for instance in sleep studies.

An extension of an actigraph-like research prototype, called Porcupine, is presented in order to be able to perform long-term recordings. The scenario of long-term monitoring for sleep requires a recording time of at least 8 hours. Development included programming of the Porcupine as well as a client which manages the Porcupine and the data on it on a client PC.

In a special experiment, performed in a sleeping lab, an idea is given on how sleep phase detection might be possible with the Porcupine. In a second experiment, data was gathered from four subjects, wearing the Porcupine at the wrist during two consecutive nights. Cluster precisions is used to identify tilt switch patterns that are allocated to body postures. Experiments show that with the Porcupine a high distinction between body postures can be achieved.

Marko Borazio

Examiner: Prof. Bernt Schiele

Co-examiner/Advisor: Dr. Kristof Van Laerhoven

Download (pdf)