Accession Number : ADP006078

Title :   The Application of Non-Stationary Data Analysis Techniques in the Identification of Changes in the Electroencephalogram Associated with the Onset of Drowsiness,

Corporate Author : ROYAL AIR FORCE FARNBOROUGH (UNITED KINGDOM) INST OF AVIATION MEDICINE

Personal Author(s) : Wright, Nicola A. ; Borland, R. G. ; McGown, Amanda S.

Report Date : FEB 1988

Pagination or Media Count : 5

Abstract : The electrical activity of the brain was analysed using techniques to detect the occurrence of non-stationarities associated with transitional states between alert wakefulness and sleep. Eight minutes of resting eyes closed data were used in the analysis. A visual inspection was carried out to classify the record into epochs of varying lengths according to the different states of arousal. Three states were defined, alert wakefulness, drowsy sleep and a transitional state. Non-stationary data analysis techniques were used to identify these changes automatically. The techniques used were autoregressive modelling, in which the prediction error was used as a criterion to detect change, and evolutionary power spectrum analysis, where a spectral ratio was defined to detect differences between short epochs of the signal. In addition, the autocorrelation function was calculated for a limited number of lags, and changes in the function with reference to previous epochs used to identify the onset of change. (js)

Descriptors :   *BRAIN, AUTOCORRELATION, DATA PROCESSING, ELECTRICAL PROPERTIES, ELECTROENCEPHALOGRAPHY, ERRORS, EVOLUTION(GENERAL), EYE, FUNCTIONS, IDENTIFICATION, METHODOLOGY, POWER SPECTRA, PREDICTIONS, RATIOS, SLEEP, SPECTRA, SPECTRUM ANALYSIS, TRANSITIONS, VISUAL INSPECTION.

Subject Categories : Psychology
      Medicine and Medical Research

Distribution Statement : APPROVED FOR PUBLIC RELEASE