Deep Learning of Biomechanical Dynamics in Mobile Daily Activity and Fall Risk Monitoring

Abstract

Daily activity and fall risk monitoring is highly important. There are 7 million fall injuries per year, and appropriate exercise can lower the risk of death by up to 20 to 70%. However, it is very challenging to accurately identify an activity due to the diversity of the human biomechanical dynamics. We propose a new intelligent computational approach, leveraging biomechanical dynamics enhancement and deep learning technologies. The detection accuracy of a total of 11,770 activities and 17 activity types is as high as 93.9%. This research is expected to greatly advance mobile activity monitoring in smart health era.