Electro-oculography-based detection of sleep-wake in sleep apnea patients

INTRODUCTION: Recently, we have developed a simple method that uses two electro-oculography (EOG) electrodes for the automatic scoring of sleep-wake in normal subjects. In this study, we investigated the usefulness of this method on 284 consecutive patients referred for a suspicion of sleep apnea who underwent a polysomnography (PSG).

METHOD: We applied the AASM 2007 scoring rules. A simple automatic sleep-wake classification algorithm based on 18-45 Hz beta power was applied to the calculated bipolar EOG channel and was compared to standard polysomnography. Epoch by epoch agreement was evaluated.

RESULT: Eighteen patients were excluded due to poor EOG quality. One hundred fifty-eight males and 108 females were studied, their mean age was 48 (range 17-89) years, apnea-hypopnea index 13 (range 0-96) /h, BMI 29 (range 17-52) kg/m(2), and sleep efficiency 78 (range 0-98) %. The mean agreement in sleep-wake states between EOG and PSG was 85% and the Cohen's kappa was 0.56. Overall epoch-by-epoch agreement was 85%, and the Cohen's kappa was 0.57 with positive predictive value of 91% and negative predictive value of 65%.

CONCLUSIONS: The EOG method can be applied to patients referred for suspicion of sleep apnea to indicate the sleep-wake state.