Hierarchical Deep Learning for Enhanced Parkinson’s disease Detection via Handwriting Analysis

Parkinson’s disease (PD) is a progressive neurological disorder that affects motor functions such as tremor, rigidity, and impaired handwriting. Early detection of Parkinson’s disease is essential for timely treatment and effective patient care. This paper proposes a hierarchical deep learning framework for detecting Parkinson’s disease using handwriting analysis. The proposed system analyzes handwriting characteristics such as tremor intensity, stroke formation, writing pressure, and spatial irregularities. A deep learning model based on MobileNetV2 is used to automatically extract relevant handwriting features from input images. The hierarchical architecture first recognizes the drawing or handwriting pattern and then performs Parkinson’s disease classification. Experimental evaluation on the NewHandPD dataset demonstrates that the proposed model improves classification accuracy compared with traditional machine learning techniques.