An enhanced Runge Kutta boosted machine learning framework for medical diagnosis

Area of Science:

• Computer Science
• Artificial Intelligence
• Medical Informatics

Background:

• Machine learning methods are increasingly used for medical diagnosis, assisting clinicians in patient care.
• The performance of machine learning models, such as Kernel Extreme Learning Machine (KELM) and Residual Neural Networks (ResNet), is highly sensitive to hyperparameter settings.
• Optimizing hyperparameters is crucial for enhancing the accuracy and robustness of machine learning-based medical diagnostic tools.

Purpose of the Study:

• To propose an enhanced optimization algorithm, GORUN (Grey Wolf Optimizer with Runge Kutta), for adaptively tuning machine learning hyperparameters.
• To improve the performance of machine learning models used in medical diagnosis by addressing the limitations of existing optimization methods.
• To develop a robust machine learning framework for medical diagnosis through effective hyperparameter optimization.

Main Methods:

• Development of the GORUN algorithm, integrating grey wolf and orthogonal learning mechanisms to enhance the Runge Kutta optimizer.
• Validation of GORUN's performance against established optimizers using IEEE CEC 2017 benchmark functions.
• Application of GORUN to optimize hyperparameters for KELM and ResNet models in medical diagnostic tasks.

Main Results:

• GORUN demonstrated superior performance compared to other optimizers on benchmark functions.
• The GORUN-optimized KELM and ResNet models achieved improved performance in medical diagnosis tasks.
• Experimental validation on multiple medical datasets confirmed the effectiveness of the proposed machine learning framework.

Conclusions:

• The GORUN algorithm offers a significant improvement over existing optimizers for complex optimization problems.
• Applying GORUN for hyperparameter tuning leads to more robust and accurate machine learning models for medical diagnosis.
• The proposed GORUN-based framework shows great potential for advancing AI-driven medical diagnostic systems.