DeepFOG: Detecting Freezing of Gait with IMUs

DeepFOG is a deep learning model for detecting freezing of gait (FOG) using IMUs. This GitHub repository contains de-identified IMU walking data from seven subjects with Parkinson’s disease in in-clinic experiments. It also contains code to pre-process, train, and evaluate convolutional neural networks that identify FOG in continuous inertial measurement unit data. The intent of this repository is to provide other researchers tools to develop and train their own FOG detection models. The paper was published the Journal of NeuroEngineering and Rehabilitation in February 2022.