Abstract

Parkinson’s disease (PD) is a progressive neurological disorder affecting movement and speech, necessitating early detection for better treatment outcomes. This research proposes a machine learningbased approach for PD detection using both image and speech data, including MRI scans, spiral drawings, and speech recordings from diagnosed individuals and healthy controls. Feature extraction techniques such as Mel-Frequency Cepstral Coefficients (MFCC) for speech and texture-based analysis for images are applied to enhance model performance. The dataset undergoes preprocessing to remove noise and standardize features before classification using Support Vector Machine (SVM), Random Forest (RF), and Decision Tree (DT). Model evaluation based on accuracy, precision, recall, and F1- score reveals that SVM achieves the highest accuracy, followed by RF and DT. Results indicate that integrating multimodal data improves PD detection accuracy, offering a non-invasive and cost-effective diagnostic solution. This study contributes to AI-driven medical diagnostics and paves the way for future research incorporating deep learning for enhanced detection.