Detecting Early Signs of Parkinson’s Disease: A Machine Learning-based Approach for Diagnostic Assistance

Parkinson’s disease (PD) is a neurodegenerative disorder resulting from the impairment or death of dopamine-producing nerve cells in the brain, primarily affecting movement. This paper presents a comprehensive approach for Parkinson disease detection through the integration of voice and handwriting analysis. Leveraging Convolutional Neural Networks for handwriting detection and Principal Component Analysis for voice analysis, the proposed method aims to accurately identify individuals with Parkinson disease. Handwriting traces and templates are compared, while voice features are extracted and analyzed to capture unique characteristics associated with Parkinson disease. Experimental results demonstrate promising accuracy, with Convolutional Neural Network achieving 87% accuracy in handwriting detection and K-Nearest Neighbour’s revealing distinct voice patterns indicative of Parkinson disease with $96 \%$ accuracy. By combining these two modalities, this interdisciplinary approach offers a more robust and reliable method for Parkinson disease diagnosis. Detecting Parkinson disease at an early stage is vital as it enables the prompt implementation of interventions and tailored treatment plans, leading to notable enhancements in patient prognosis. Moreover, the proposed method addresses the limitations of traditional diagnostic techniques by providing a non-invasive and cost-effective solution. This research significantly contributes to the progression of methodologies for detecting Parkinson’s disease, thereby facilitating the development of diagnostic tools that are both more accessible and precise for use in clinical settings.