A model-based approach to homeostatic and ultradian aspects of nocturnal sleep structure in narcolepsy.

STUDY OBJECTIVE

We present a mathematical model of sleep-EEG structure applied to the analysis of sleep patterns in narcoleptics by combining the 2-process model of sleep regulation and the reciprocal interaction model of REM regulation suggested by McCarley and Hobson. The aim was the individuation of parameters characterizing narcoleptic sleep in comparison to controls.

DESIGN

Polysomnographic data were drawn from a previous study about sleep in narcolepsy. The mathematical model was fitted to quantitative EEG data by an optimization procedure.

SETTING

Polysomnographic data were recorded in single and sound attenuated hospital rooms, for one night following an adaptation night.

PARTICIPANTS

9 narcoleptic subjects (7 males, 2 females, mean age 39.6 +/- 4.3 years) and 9 age- and sex- matched controls.

MEASUREMENTS

Slow Wave Activity (SWA) time series were evaluated by spectral analysis. The sleep model was fitted to SWA profile for each recording and to the averaged SWA profile for each group. Bartlett and Kolmogorov-Smirnov test were used to evaluate the goodness of fit and the accuracy of model predictions.

RESULTS

In both controls and narcoleptics the optimization procedure produced a good fit of SWA raw data. The only significant difference between the groups were the RemOn/RemOff coupling parameters, reflecting an enhanced strength of the REM oscillator in narcoleptics.

CONCLUSIONS

The mathematical model of sleep provides a substantial description of empirical patterns for both controls and narcoleptics. The variation of values in the parameters describing the strength of RemOn /RemOff interaction is the major feature characterizing narcoleptics; it can explain sleep onset REM periods (SOREMPs) and variations of REM-NREM sleep cycle duration.