TremTrack: Mobile Classification of Parkinsonian Hand Tremors Using Accelerometer Data

Parkinson’s Disease (PD), affecting 10 million people worldwide, is the second most common neurodegenerative disease. PD often starts with uncontrollable hand movements called tremors which hinder normal daily activities. PD does not have a cure, meaning early detection is crucial. This study utilized the Balanced Random Forest classifier (87% accuracy, $\mathbf{0. 9 3}$ AUC) and data from the PD-BioStampRC21 dataset to make a mobile app that classifies Parkinson’s in free-living conditions using methods from activity and PD classification using accelerometer data. The app is designed to be accessible to the general public, aiming to provide earlier detection of PD.