USING MACHINE LEARNING TO DETECT PARKINSON'S DISEASE

Parkinson's disease is a progressive neurological disorder that affects millions of people worldwide. Early and accurate detection of this disease is crucial in managing its progression and improving patient outcomes. In this study, machine learning algorithms were applied to a dataset containing 195 instances and 24 features representing both healthy individuals and those diagnosed with Parkinson's disease. The analysis showed that the dataset was unbalanced. Approximately 75% of the records corresponded to individuals diagnosed with Parkinson's disease. Five algorithms were implemented to evaluate the effectiveness of machine learning models in detecting Parkinson's disease: Logistic Regression, Support Vector Machine (SVM) with Linear Kernel, Decision Tree, Random Forest, and K-Nearest Neighbor (KNN). The results showed that the Random Forest and KNN models achieved superior performance compared to other methods. In particular, the KNN model showed the highest accuracy (95%), precision (94%), recall (100%), and F1 score (97%). The Random Forest model also achieved high performance with accuracy (92%) and other metrics close to KNN, indicating its reliability for this classification task. On the other hand, Logistic Regression and SVM showed mediocre results, with precision, recall, and F1-scores below 95%. This study contributes to the growing field of medical diagnostics by demonstrating the potential of machine learning methods to detect Parkinson's disease.