Parkinson's disease prediction using improved crayfish optimization based hybrid deep learning.

Area of Science:

• Neurology
• Biomedical Engineering
• Artificial Intelligence

Background:

• Accurate prediction of Parkinson's disease progression is crucial for timely diagnosis and treatment, ultimately improving patient outcomes.
• Current prediction methods may benefit from novel approaches leveraging accessible resources and advanced computational techniques.

Purpose of the Study:

• To develop and validate a novel, freely accessible method for Parkinson's disease prediction.
• To enhance feature extraction and classification accuracy using advanced signal processing and optimization algorithms.

Main Methods:

• Data preprocessing involved band-pass filtering, followed by Empirical Mode Decomposition (EMD) for feature extraction.
• An Attention-based Efficient Bidirectional Network (ImCfO_Attn_EffBNet), integrating EfficientNet-B7, BiLSTM, and Attention modules, was employed for classification.
• The Improved Crayfish Optimization (ImCfO) algorithm was utilized to optimize the network's convergence, loss function, and parameters.

Main Results:

• The ImCfO algorithm demonstrated enhanced performance through a self-adaptation criterion, improving convergence and finding optimal solutions.
• The ImCfO_Attn_EffBNet achieved high prediction accuracy (95.068%), with strong recall (92.948%), specificity (92.89%), and F-Score (92.89%).
• The method effectively gathered temporal and spatial data for robust Parkinson's disease prediction.

Conclusions:

• The developed ImCfO_Attn_EffBNet model shows significant promise for accurate and reliable Parkinson's disease prediction.
• The integration of EMD, ImCfO, and advanced neural network architectures offers a powerful tool for neurological disorder analysis.
• This novel approach, utilizing freely accessible resources, can aid in early diagnosis and personalized treatment strategies for Parkinson's disease.