A User-Centric Framework to Detect Parkinsons Disease

Parkinson’s Disease (PD) is a progressive neurodegenerative disorder affecting motor functions, with early detection critical for effective intervention. Traditional diagnostic methods are often invasive, expensive, and inaccessible, necessitating alternative approaches. This paper proposes a deep learning-based framework for PD detection using hand-drawn spirals and waves, leveraging Convolutional Neural Networks (CNNs) and transfer learning models (InceptionV3, Xception, EfficientNetB2, and VGG19 with attention mechanisms). We analyse digitized spiral and wave drawings from PD patients and healthy controls, employing data augmentation to enhance model robustness. Our system achieves 96.4% accuracy using EfficientNetB2 and 97.5% accuracy with VGG19-attention, outperforming traditional machine learning methods. The proposed solution offers a non-invasive, cost-effective, and scalable screening tool for early PD detection, with potential applications in telemedicine and remote diagnostics.