Early Detection of Parkinson's Disease using Motor Symptoms and Machine Learning
This work addresses early diagnosis for Parkinson's disease patients, particularly those over 60, but is incremental as it applies existing methods to new data.
This research tackled early detection of Parkinson's disease by analyzing motor and gait symptoms using machine learning on the PPMI Gait dataset, achieving a model accuracy of 91.9%.
Parkinson's disease (PD) has been found to affect 1 out of every 1000 people, being more inclined towards the population above 60 years. Leveraging wearable-systems to find accurate biomarkers for diagnosis has become the need of the hour, especially for a neurodegenerative condition like Parkinson's. This work aims at focusing on early-occurring, common symptoms, such as motor and gait related parameters to arrive at a quantitative analysis on the feasibility of an economical and a robust wearable device. A subset of the Parkinson's Progression Markers Initiative (PPMI), PPMI Gait dataset has been utilised for feature-selection after a thorough analysis with various Machine Learning algorithms. Identified influential features has then been used to test real-time data for early detection of Parkinson Syndrome, with a model accuracy of 91.9%