Neurodegenerative disorders such as Parkinson’s Disease (PD) are progressive neurological conditions that severely affect both motor and cognitive abilities, often manifesting through distinctive handwriting abnormalities. Early and accurate detection of PD is essential to enable timely therapeutic intervention and to improve patient outcomes. In this study, we propose a hybrid feature extraction framework for PD detection based on offline handwriting analysis. Our approach integrates Fourier Transform (FT), Histogram of Oriented Gradients (HOG), and Discrete Wavelet Transform (DWT) to capture both global structures and fine-grained variations in handwriting. To boost classification accuracy, a stacking ensemble model is adopted, combining multiple machine learning classifiers to exploit their complementary predictive capabilities. Experimental results on a PD-specific handwriting dataset validate the proposed method’s effectiveness, achieving an average classification accuracy of 98.49% across multiple tasks. These findings underscore the promise of offline handwriting analysis as a non-invasive, cost-efficient, and scalable diagnostic aid for neurodegenerative diseases.