A SCOPING REVIEW ABOUT RECOGNITION OF HEALTH CONDITIONS USING INERTIAL SENSOR SIGNALS

Resumo

Abstract: This scoping review mapped studies on the identification of health conditions through gait signals extracted from inertial sensors. Studies included should identify health conditions using inertial sensor signals, be available in Portuguese, Spanish or English from 2010 until 2023. The search was performed in the databases IEEE Xplore, PubMed, Scopus, Web of Science and Lilacs, in addition to gray literature search. Data were analyzed descriptively and through Bardin content analysis. 220 studies were included. The main health conditions identified using gait signals from inertial sensors were Parkinson's Disease and Multiple Sclerosis. Data were captured mainly using a sensor during an estimated path and functional tests. The main sensors used to capture gait and movement features were accelerometer, gyroscope, and magnetometer. About half of the studies (52.07%) used machine learning. The methods and variables differ greatly, and many lack clear accuracy descriptions. This review identified several health conditions that can be identified by gait signals from inertial sensors, which can promote early diagnosis, preventing its onset and progression.