ParkinSense: Deep Learning for Early Parkinson’s Detection

Abstract

Parkinson's Disease (PD) is a progressive neurological disorder that impacts millions worldwide. Early detection is essential for managing its progression and improving patient outcomes. This paper presents a graphical user interface (GUI) application designed to detect PD using machine learning (ML) techniques. The application allows users to upload biomedical voice measurement datasets, preprocess the data by handling missing values, and apply the Synthetic Minority Over-sampling Technique (SMOTE) to address class imbalance. It also includes exploratory data analysis (EDA) to visualize data distribution and relationships.The system implements six machine learning algorithms—Decision Tree, Random Forest, Logistic Regression, Support Vector Machine (SVM), Naive Bayes, and k-Nearest Neighbors (KNN)—to classify PD. Models are evaluated based on accuracy, precision, recall, and F1 score, with results displayed through confusion matrices and comparison graphs. This interactive tool offers an accessible platform for researchers, clinicians, and data scientists to efficiently explore machine learning techniques for PD diagnosis.By providing detailed performance insights, the application helps identify the most effective algorithm for specific datasets. With its user-friendly interface and comprehensive functionality, this tool supports early detection efforts, potentially enhancing patient care through more accurate and timely diagnosis of Parkinson's Disease.