An Effective Way to Improve Actigraphic Algorithm by Using Tri-axial Accelerometer in Sleep Detection

This paper presents an effective method for improving the accuracy of sleep detection using actigraphic algorithm. We have collected 2 sets of comparative data, wrist movement acceleration and brain wave analysis. In the first mode, we use the acceleration on x-axis as the input of actigraphic algorithm, in imitation of uni-axial accelerometer used in former study. In the second mode, we calculate the vector magnitude of acceleration in 3 dimensions and use it as the input of algorithm. Two outputs are then compared with Zeo data that are collected simultaneously. The comparison of results between 2 modes exhibit fine improvement on sleep detection.