Advanced PCA-KNN Classification Technique for Parkinson’s disease Diagnosis at Early Stage

Abstract

Despite the increasing clinical demand for accurate and objective methods to evaluate Parkinsonian tremors, machine learning–based scoring aligned with the Unified Parkinson’s Disease Rating Scale (UPDRS) is still underutilized. This study addresses this gap by employing machine learning algorithms to predict UPDRS scores in a way that mirrors the evaluation approach used by neurologists in clinical practice. Although traditional methods such as Bayesian Networks, Decision Trees, and Artificial Neural Networks have been applied to Parkinson’s Disease (PD) detection, there is room for improvement in terms of classification accuracy and model robustness. In this work, we propose an enhanced classification framework based on Principal Component Analysis (PCA) combined with the K-Nearest Neighbors (KNN) algorithm to improve the diagnostic accuracy of Parkinson’s disease. The proposed methodology is implemented in a Jupyter Notebook environment using Python, which provides a flexible and open-source platform for data preprocessing, model training, and performance evaluation. Li-braries such as Scikit-learn, NumPy, and Matplotlib are utilized for dimensionality reduction, classification, and visualization, respectively. Performance evaluation based on accuracy and precision demonstrates that our PCA-KNN model significantly outperforms conventional methods, highlighting its potential as a reliable and efficient diagnostic approach for Parkinson’s disease. Index Terms—Parkinson’s Disease (PD),Unified Parkin-son’s Disease Rating Scale (UPDRS),Machine Learning,PCA-KNN (Principal Component Analysis – K-Nearest Neigh-bors),Classification ,Dimensionality Reduction Python.