Handwriting Dynamic Assessment using Deep Neural Network Multi Layer Perceptron for Parkinsons Disease Classification*

Parkinson's disease (PD) is a neurodegenerative disorder that significantly affects motor functions, often leading to tremors and abnormal handwriting dynamics. Early diagnosis is crucial for effective treatment, yet accurate detection remains a challenge due to the complex nature of motor impairments. This study explores a novel application of a Deep Neural Network Multi-Layer Perceptron (DNN-MLP) model to assess handwriting dynamics and classify individuals into control or patient groups. The dataset utilized consists of various handwriting metrics such as root mean square, mean relative tremor, and differences in radii between exam templates and handwritten traces. By training the DNN-MLP on these features, an accuracy of 83% was achieved, surpassing other traditional classifiers such as Support Vector Machine (81%), Random Forest (75%), and k-Nearest Neighbours (77%). Our results demonstrate that the DNN-MLP model's ability to capture intricate handwriting patterns makes it a promising tool for the early detection of Parkinson's disease. Further refinement and larger datasets could enable the deployment of this approach in clinical settings, potentially improving diagnostic accuracy and patient outcomes.