Parkinson's disease (PD) is a progressive neurodegenerative disorder that primarily affects motor function, leading to tremors, stiffness, and difficulty with movement and coordination. Early detection of PD is critical for timely intervention and improved patient outcomes. This study introduces a novel method for PD detection through the analysis of activity-related data using a Transformer-based model. The Transformer architecture, renowned for its ability to handle sequential data, leverages self-attention mechanisms to capture complex temporal dependencies and patterns. Unlike traditional models such as LSTM (Long Short-Term Memory), which processes data sequentially, Transformers enable parallel processing, enhancing computational efficiency and boosting prediction accuracy. This approach bypasses the need for image-based analysis and instead focuses on extracting meaningful insights from activity-related data, making it a more accessible and cost-effective solution for early PD detection. The proposed method not only enhances prediction accuracy but also provides a non-invasive and user-friendly tool for PD screening. Preliminary results demonstrate that the proposed hybrid framework achieves a classification accuracy of 91.5%, outperforming traditional models such as SVM and Random Forest. The findings emphasize the potential of deep learning in advancing neurodegenerative disease diagnostics. Future research will focus on refining the model and expanding the dataset to improve its robustness and generalizability across diverse populations.