Mapping Raw Acceleration Data on ActiGraph Counts: A Machine Learning Approach

Abstract

A method for mapping actimetry data between two platforms has been carried out; one platform is the CE and FDA approved ActiGraph wGT3X-BT, valid in clinical diagnosis, and the second one is Microsoft Band 2 Smartband, available for the general public. The method consists of a regression performed using machine learning technique, specifically, different configurations of neural networks have been tested. Access to the data has been achieved by means of an in-house mobile application.