Parkinson's Disease Detection: A Comparative Analysis of Voice- and Spiral/Wave Drawing based Detection using Machine Learning Techniques

Parkinson’s Disease (PD) is a gradually worsening brain disorder that mainly affects movement and speech. Catching it early is really important so we can start treatment sooner and try to slow down how quickly it gets worse. In this study, we introduce a new approach that uses machine learning to help detect PD early. We look at three different types of data: drawings of spirals, wave patterns, and recordings of voices. Each type of data goes through its own process to find key features—like irregularities in drawings or voice qualities such as jitter, shimmer, and the ratio of harmonic to noise. We then use combination classifiers, specifically Random Forest and XGBoost, to analyze the data. Results show that the spiral drawing method was the most accurate, with about 88% success, followed by the wave pattern with around 85%, and voice recordings at roughly 76%. These results match previous research that shows motor tasks like drawing can be good early signs of PD. Because this system is modular and doesn’t require invasive testing, it has a lot of potential for remote healthcare and telemedicine. It offers an affordable and easy way to screen for PD early on so that’s accessible to many people.