An Optimization Method for Classifying Parkinson's Disease

A quick and correct diagnosis is necessary for effective treatment of Parkinson's disease. In this study, we employ a comprehensive feature optimization methodology and cutting-edge machine learning algorithms to enhance the classification accuracy of Parkinson's disease. Our chosen dataset comprises 756 occurrences and 754 features, giving our study a solid foundation. This study efficiently reduces the dimensionality while preserving crucial information by identifying and selecting the top 40 most pertinent features using the information gain feature reduction technique. Based on this optimized feature set, it is then used to train and assess a variety of classifiers, like AdaBoost, the Random Forest model, Support Vector Machine (SVM), and Logistic Regression. Among these, the Logistic Regression classifier demonstrated the highest accuracy, achieving an impressive 81.5%. Our results underscore the effectiveness of the proposed feature optimization method in improving classification performance. Besides enhancing early diagnosis of PD, this approach could also improve patient outcomes and quality of life by developing personalized treatment strategies. This research highlights the importance of feature selection in medical diagnostics and its role in advancing predictive accuracy.