Machine Learning Based Parkinson’s Disease Detection Using Voice and Handwriting Analysis

This paper is based on detection of Parkinson disease using machine learning models accurately. In this paper two major symptoms of Parkinson are taken into consideration one is shimmering of voice and another is tremor in hands. They are applied with sufficient datasets two obtain accurate results. For Shimmering in voice we visualized the data and divided the dataset in 23 unique features and then we applied Extreme Gradient Boosting algorithm (XGBoost) to train the model and obtained an accuracy of 92.3%, and for tremors in hands we collected dataset of patients with drawing of spirals and waves, then we passed it through feature processing and used Histogram of oriented gradient, to obtain features and then attached them with tags of Parkinson and Non Parkinson, then passed it to random forest classifier algorithm to train and obtain an accuracy of 82%. We also obtained results for real time detection and data set management, using Praat software and able to extract features of voice on real time.