Egocentric Vision and IMU Fusion for Parkinson's Freezing of Gait Detection

An IMU-based Temporal Convolutional Network (TCN) achieved 42.3 F1 and 83.0 AUROC in detecting freezing of gait for Parkinson's patients, outperforming frozen representations from pretrained ego-video and time-series foundation models. The study uses synchronized egocentric video and wearable IMUs to disambiguate pathological movement from intentional activities of daily living.