Measuring Sleep by Wrist Actigraph
Abstract
A convenient method of monitoring personnel sleep and activity in field conditions is needed to promote medical planning for modern combat. In the period from April 1980-March 1981, we have programmed, tested, and begun evaluations of a wearable digital activity system, and we have refined a computer process for recognizing sleep from this system. Together, these efforts enable us to collect data from freely ambulatory subjects which can be scored automatically for sleep/wake with accuracy comparable to EEG scoring. The system is ready for miniaturization leading to field use. Our microprocessor-based digital activity monitor was built to our specifications, and we added external activity and illumination transducers, Actual data collection was implemented and 25 records totaling over 27,000 minutes have been obtained as of March, 1981. Fourteen records (12,739 minutes) collected with the digital monitor were scored retrospectively with 93.6% agreement with EEG sleep/wake scoring. Research is continuing to further increase the accuracy of the sleep recognition algorithm. Since the errors that occur include both mis-scoring sleep as wake and vice versa, they tend to cancel. Correlations between sleep durations scored from activity data and from EEG records were r=0.9760 for digital monitor data. Current results now allow us to specify design criteria for a miniaturized wrist-mounted activity monitor suitable for field or combat use.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1981
- Accession Number
- ADA103196
Entities
People
- Daniel F. Kripke
- Daniel J. Mullaney
- John B. Webster
- Sam Messin
- William Mason
Organizations
- University of California, San Diego