Detection of Parkinson's Disease by Using Machine Learning Stacking and Ensemble Method

Using speech data, it is difficult to learn through machine learning how diagnose Parkinson's disease (PD) and evaluate the effects of treatment. For this issue, article has developed a three-stage PD discovery method. The base classifiers used in initial stage are logistic regression (LR), K-nearest neighbor (KNN), naive bayes (NB), support vector (SVC), decision tree (DT). second stage, or stack model, meta-model that combines all mentioned earlier. third ensemble model consists Bagging, AdaBoost, Random Forest (RF), Gradient Boosting (GBC) components. RF GBC utilized estimate most important features from dataset. models' validation been evaluated using confusion matrix metrics like precision, recall, F1 score. Out models, GBC—the model—had highest accuracy with testing data—97.43%. KNN stacking meta-model, on other hand, had accuracy, 94.87% each. models manuscript, only classifier accuracy. proposed appears be an extremely useful for disease, as demonstrated by exploratory findings factual analyses.