Sleep estimation from wrist movement quantified by different actigraphic modalities

Progress in transducer design and empirical characterization of wrist movem ent has led to diverse wrist activity monitors, each with its unique featur es and modality of operation. This study compared sleep-wake estimates from nocturnal wrist activity quantified by different motion-quantifying algori thms. Healthy young adults wore an Actillume and a Mini Motionlogger on the same wrist while nocturnal polysomnography data were recorded simultaneous ly in the laboratory. Activity data were analyzed with ACTION3 using scorin g algorithms independently calibrated for each measurement modality. Overal l, each modality yielded accurate and reliable sleep estimates relative to polysomnographic estimates (agreement rates: 91.4-96.5%, correlations for s leep duration: 0.79-0.94). Estimates derived from Actillume modalities were comparable to those of Milli Motionloggers, suggesting that the transducer s of these two devices performed comparably for monitoring sleep and wakefu lness. Wrist movement quantified by the Mini Motionlogger proportional-inte grating mode yielded the best accuracy for detection of sleep-wake states. (C) 2001 Elsevier Science B.V. All rights reserved.