Fig. 1: Study design and data modeling setup.
From: Modeling multiple sclerosis using mobile and wearable sensor data
We ran the experiments on a dataset we collected from 55 PwMS and 24 healthy controls over two weeks. Our data analysis objectives are to identify the most reliable, clinically useful, and available features derived from mobile and wearable sensor data. Our machine learning pipeline identifies the best-performing features for four tasks 1) PwMS and healthy controls classification, 2) MS type classification into none, RRMS and PPMS-SPMS, 3) predict the disability and fatigue levels using wearable sensor data, smartphone usage, and demographics. [Illustrations by Storyset (https://storyset.com/)].
