Sleep-Wakefulness Determinations from Heart Rate Data,
Annual rept. 1 May 75-30 Mar 76,
TEXAS UNIV AT AUSTIN
Pagination or Media Count:
During the second year of this research the REM-NREM algorithm developed during the first year of work to separate the sleep cycles on the basis of minute-by-minute heart rate was re-evaluated on new data as was the algorithm for separation and classification of data into four stages Awake Stage 0, Stage 15, Stage 2, and Stage 34. Development of two discriminant analysis algorithms, one for REM-NREM separation and the other for a four stage separation was also begun at this time. Several nights of data were used individually as training sets with each type of analysis. For the REM-NREM classification based on the first years algorithm the training night accuracies ranged from 58.3 to 83.0. For the REM-NREM classification based on the new algorithm, the training night accuracies ranged from 87.3 to 89.2. Results from the new discriminant analysis four-stage algorithm, yielded accuracies ranging from 85.0 to 93.2 on one subject.
- Anatomy and Physiology