Intelligent Parkinson's disease identification via Residual-Shuffle Network optimized by Improved Dandelion Optimizer

Area of Science:

• Neurology
• Computer Science
• Artificial Intelligence

Background:

• Parkinson's disease (PD) diagnosis is often delayed, hindering effective management.
• Early detection is crucial for slowing disease progression and improving patient outcomes.
• Machine learning (ML) methods show promise for automated PD detection.

Purpose of the Study:

• To develop an accessible and precise ML-based automated system for early Parkinson's disease detection.
• To combine the Residual-Shuffle Network (ResNet) with the Improved Dandelion Optimizer (IDO) for enhanced PD identification.

Main Methods:

• A novel framework, IDO-ResShuffle, was proposed, integrating ResNet with IDO.
• IDO features adaptive parameter control and balanced exploration-exploitation for optimizing hyperparameters and network weights.
• The model was evaluated on the HandPD dataset for handwriting pattern analysis.

Main Results:

• The IDO-ResShuffle framework achieved high performance metrics: 97.6% accuracy, 96.9% F1-score, 97.2% sensitivity, and 97.1% specificity.
• The joint optimization of network architecture and hyperparameters by IDO proved effective for PD detection.
• The method demonstrated reliable identification of Parkinson's disease from handwriting without complex data collection.

Conclusions:

• The proposed IDO-ResShuffle framework offers a reliable and accessible solution for automated Parkinson's disease detection.
• This approach can empower healthcare professionals with earlier diagnostic information for timely intervention.
• Enhanced clinical decision-making and support for Parkinson's disease management are facilitated by this AI-driven method.