Generalizing and Mapping Actigraphy Based Sleep Detection Algorithms

Numerous actigraphy sleep detection algorithms have been developed and thoroughly evaluated against clinical polysomnogram (PSG) sleep studies. These widely used actigraphy sleep detection algorithms have been developed with non-standard actigraphy-counts, which were specific to the devices used to collect the actigraphy data. The non-standard actigraphy-counts have made the comparison and reproduction of many sleep studies difficult and error-prone, leading to varying rescoring approaches to improve accuracy. Herein we present an alternate method to map common actigraphy-count sleep detection algorithm coefficients. The mapped algorithms use a standard vector magnitude gravity (VMG) from a triaxial accelerometer in standard gravity units. In addition to mapping the inputs, the algorithms are also generalized to use configurable epochs. The mapped actigraphy algorithms will allow researchers and clinicians to leverage emerging devices that considerably extend runtimes, and provide consistent results across various device types by using a standardized actigraphy units of measurement.