Freezing of Gait Prognostication in Parkinson’s Disease

Parkinson’s disease (PD) affects millions worldwide, and a significant portion experience freezing of gait (FOG), a disabling symptom that impedes mobility and increases fall risk. Despite extensive research, the underlying mechanisms of FOG and effective treatment strategies remain elusive. Objective and precise FOG detection and classification are crucial for advancing our understanding and management of this symptom. Existing FOG detection methods face limitations in accuracy, generalizability, and the ability to distinguish between FOG subtypes. Moreover, current treatment options for FOG are limited and often provide suboptimal outcomes. To address these challenges, we propose a novel approach utilizing machine learning and wearable sensor data to accurately detect and classify FOG episodes. We employ a comprehensive dataset comprising 3D accelerometer data from the lower back of FOG subjects. Using advanced machine learning models, we aim to identify the onset and cessation of FOG episodes and classify them into three distinct types: Start Hesitation, Turn, and Walking. This approach holds the potential to overcome the limitations of existing methods and provide a more comprehensive understanding of FOG. Our research aims to shed light on the intricate mechanisms of FOG, paving the way for the development of more effective treatments and improved quality of life for individuals living with Parkinson’s disease.