Pattern Recognition of EEG to Determine Level of Alertness

Abstract

The report documents the work accomplished during the fourth reporting period in applying the principles of pattern recognition technology to the analysis of EEG. Several additional sleep scoring decision systems were investigated. Most classifications are performed satisfactorily. However, no reliable single channel frequency relationships have been discovered for separation of Stage 1 and REM. Also the transitional nature of Stage 3 patterns leads to confusing these patterns with Stage 2 or Stage 4. Cluster analysis was applied to frequency indicators derived from an overnight sleep record. Selection of clusters independent of human scoring demonstrated well defined groupings for awake, Stage 1 and REM combined, Stage 2 and Stage 4. Stage 3 patterns were split between the Stage 2 and Stage 4 clusters. Sleep scoring decision systems based upon frequency analysis of a single channel of EEG have been designed for each of two Tektite I aquanauts.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jun 01, 1970
Accession Number
AD0726473

Entities

People

  • William B. Martin

Tags

Communities of Interest

  • Human Systems

DTIC Thesaurus Topics

  • Classification
  • Clustering
  • Data Processing
  • Databases
  • Frequency
  • Human Factors Engineering
  • Indicators
  • Magnetic Tape
  • Military Research
  • Neurobehavioral Manifestations
  • Pattern Recognition
  • Physiological Phenomena
  • Recognition
  • Recovery
  • Sleep Deprivation
  • Spectra
  • Standards

Readers

  • Circadian Sleep-Wake Regulation and Chronobiology
  • Regression Analysis.
  • Theoretical Analysis.

Technology Areas

  • AI & ML