Screening of sleep architecture using a disposable patch sensor

Tremendous interest prevails in developing a convenient and effective wearable medical device for evaluating sleep at home. The current study presents the clinical validation of VitalPatch®, a wireless adhesive patch sensor for screening of sleep architecture compared to the gold standard polysomnograph (PSG). A group of 45 volunteers were attached to a standard 22-channel PSG and a VitalPatch sensor on their chest. Simultaneous PSG and patch data were acquired overnight wirelessly. Hypnograms with wake, non-rapid eye movement (NREM), and REM stages were obtained using both the methods for a sequence of 30 s epochs corresponding to the bed-time, and the respective statistical sleep metrics were computed in 42 subjects. Performance, statistical and agreement analyses were conducted using VitalPatch's sleep assessments compared to the PSG. The accuracy and Cohen's kappa measures of 3-class sleep stage prediction (n=42) was 80.5±8.3% and 0.50±0.18, respectively. Total sleep time showed a strong linear correlation (r=0.88) with a mean absolute error (MAE) of 25 min compared to the PSG. NREM and REM sleep times were highly correlated to the PSG (r=0.82 and r=0.71, respectively) with MAE of 26 min and 16 min, respectively. The results suggest that unobtrusive disposable VitalPatch sensor is effective in predicting sleep architectures highly comparable to the PSG and can be useful for widespread screening of sleep patterns in home environment.