Optimized technique for the early identification of Parkinson's disease using machine learning-based handwriting and voice analysis

Abstract

Movement impairments caused by Parkinson's disease include tremors and rigidity in the muscles. Due to the fact that each patient's symptoms are unique, PD can easily go undiagnosed. Because of this, the early phases of Parkinson's disease, when symptoms are modest, are not well diagnosed with existing procedures. New research, however, supports the idea that PD may be identified early by using handwriting and speech abnormalities as indicators. Early detection of Parkinson's disease is essential since it guarantees that the illness's initial occurrence is identified and treated as soon as feasible. In addition, this method allows doctors to better control how the illness progresses. In the end, patients are able to prolong their lives with their loved ones. However, early PD diagnosis and the required early treatments are made possible by data-driven identification models. Researchers can improve patient quality of life and speed up therapy by optimizing diagnostic accuracy with data analytics and technology. In this paper, proposed work different machine learning algorithms like Support Vector Machine, Decision Tree, Random Forest, K Nearest Neighbor algorithms are applied, in that SVM and Random Forest show better performance for three models, yielding accuracy levels of 97%, 86.67%, and 76.65% for the Voice, Spiral, and Wave models, respectively.