Multi Class Parkinson Disease Detection Based on Finger Tapping Using Attention Enhanced CNN BiLSTM

The recent study on multi-class Parkinson's disease detection highlights the importance of accurately assessing PD severity for better clinical management. Traditional gesture-based recognition systems, including those using finger tapping, have struggled with performance, prompting the need for innovative solutions. This study introduces an attention-enhanced CNN BiLSTM framework that processes an existing dataset of finger tapping videos to extract critical features related to wrist and hand movements. By employing a hybrid deep learning approach, the model classifies PD severity into multiple levels, aiming to provide a more reliable assessment tool. The integration of handcrafted feature extraction with advanced deep learning techniques represents a significant step forward in addressing the challenges faced by existing systems, ultimately contributing to improved patient care and intervention strategies.