Freezing-of-Gait Detection Using Wearable-Sensor Technology and Neural-Network Classifier

Freezing of gait (FoG) is a syndrome of Parkinson's Disease (PD), which causes patients to be unable to initiate and/or maintain the rhythmical and alternating leg movements of locomotion. Since FoG negatively impacts patient quality of life and is a main reason for PD patients to fall, the ability to quantify FoG episodes with reliable and accurate algorithms could help enable high-value therapies to be developed. In this study, a wearable sensor platform is used to collect inertial and physiological signals to cover all the variations of freezing patterns. A diverse set of features is also introduced and fed to a Fully-Connected-Neural Network (FCNN) to learn and predict freezing episodes.