Using Machine Learning Algorithms to Predict Parkinson's Disease : A Comparative Analysis Based on Logistic Regression, Random Forest and XGBoost

Abstract

Parkinson's disease, as a neurodegenerative disease, seriously affects the quality of life of patients. Early and accurate diagnosis is crucial for treatment. For comparison and prediction, this study makes use of the Extreme Gradient Boosting (XGBoost) method, random forest, and logistic regression. Data preprocessing includes removing irrelevant features, deleting duplicate records, processing missing values, and Z-score standardization. The results show that the XGBoost model performs best on the test set, with an recall rate, accuracy rate, precision rate and F1 score of 0.87, which is greater than logistic regression (0.774) and random forest (0.862). The AUC value of XGBoost is 0.95, and the classification ability is the best. Despite overfitting, XGBoost performs well in dealing with complex nonlinear relationships. Future research can optimize hyperparameters, expand data sets, and explore deep learning techniques to improve prediction accuracy and generalization ability further. This research offers evidence in favor of machine learning-based Parkinson's disease prediction, which is crucial for raising the bar for early detection.