Clustering of sleep electroencephalographic patterns in patients with the fibromyalgia syndrome

Several electroencephalographic (EEG) abnormalities have been observed during sleep in patients suffering from the fibromyalgia syndrome (FMS). In this study, 12 patients with fibromyalgia and 14 control subjects had two polysomnographic recordings obtained at home. Data from the second night were subjected to blinded manual scoring as well as signal processing using linked or 'step-wise clustering for pattern recognition. In this procedure, a common learning set was generated using the spectral information in three 2 min EEG samples from each of the sleep stages selected from five patients with FMS and five controls. In this way, 17 characteristic EEG classes were defined. All 2 s EEG segments from the whole night from all subjects were then assigned to one of these classes. Five of the classes (dominated by 0.5-4.5 Hz activity) were more frequent in the control group, whereas three other classes (dominated by 8-11 Hz activity) were prevalent in the patient group. This trend was consistent in all sleep stages, although most striking in non-rapid eye movement (NREM) sleep. The predominance of these classes in the patient group may correspond to the alpha-EEG sleep anomaly previously reported in subjects with FMS. More importantly, as the EEG power in the lowest frequency range (prevalent in controls) probably is a marker for restorative sleep, the findings may reflect important aspects of sleep disturbances n subjects suffering from FMS, thereby contributing to some of the daytime symptoms in these patients.