Machine Learning and Deep Learning Models for Diagnosis of Parkinson’s Disease: A Performance Analysis

Abstract

Parkinson’s disease (PD) is a complex condition that is characterized by restricted mobility. Symptoms begin gradually, with only one hand exhibiting a minor tremor on occasion. Also, in the beginning stages of Parkinson's disease, your face may be expressionless. The fingers are not going to vibrate. Your voice may also become mute or slurred. Parkinson's disease indications and symptoms worsen with time. The focus of this thesis is to assess the efficacy of deep learning and machine learning strategies in discovering the best and most accurate strategy for early Parkinson's disease diagnosis utilising a vast dataset from the UCI machine learning repository of 5876 × 22 fields, which includes Parkinson's and healthy people details. Performance analysis of each method is done by considering the metrics like Precision, Recall, F1-Score, Support, Confusion Matrix, Specificity and Sensitivity and are plotted in graph showing training loss and accuracy. The highest accuracy of 97.43% is achieved for KNN with k=5 (K-Nearest Neighbors) algorithm which is a supervised machine learning approach.