parkinson disease classification based on the modified binary pso and machine learning model

Parkinson’s disease is one of the most prevalent neurological disorders among the elderly. accurate diagnosis of parkinson’s disease has been shown to enhance treatment effectiveness and improve patients’ quality of life. this study presents an enhanced classification framework for parkinson’s disease by combining modified binary particle swarm optimization (mbpso) with k-nearest neighbor (knn) through the leave-one-out cross-validation (loocv). specifically, four transfer functions are employed to convert the continuous search space of pso into a binary one. additionally, the modified bpso incorporates chaotic maps and the catfish effect to enhance exploration capabilities in identifying a relevant subset of features to build a predictive model. the proposed framework is rigorously evaluated using key metrics, including accuracy, precision, recall, and f1-score. by reducing the number of features, mbpso improves both model efficiency and predictive performance. the best-performing transfer function variant is v3, which achieved an accuracy of 0.975, precision of 0.975, recall of 0.987, and f1-score of 0.981 on the training set, and obtained an accuracy of 0.935, precision of 0.967, recall of 0.95, and f1-score of 0.958 on the testing set.