A novel approach to venous clinical severity score prediction: combining metaheuristic algorithm and random forest...

Area of Science:

• Vascular Medicine
• Medical Informatics
• Machine Learning

Background:

• Varicose veins result from venous valve failure, often inadequately addressed by conventional treatments.
• Lifestyle and dietary modifications, including yoga, show potential in managing and preventing varicose veins.

Purpose of the Study:

• To develop and evaluate a machine learning model for predicting the Venous Clinical Severity Score (VCSS).
• To assess the efficacy of optimization algorithms in enhancing model performance for chronic venous insufficiency classification.

Main Methods:

• Random Forest Classification (RFC) was employed to predict VCSS across four severity categories.
• The performance of RFC was enhanced using Black Widow (BW) and Improved Artificial Optimizer (IAO) algorithms.
• Model efficacy was evaluated using precision metrics for each VCSS category.

Main Results:

• The RFBW hybrid model demonstrated superior accuracy and reliability in predicting VCSS.
• High precision scores (0.917 to 1.000) were achieved by the RFBW model across all severity levels.
• The RFIA model yielded comparable results to the RFBW model, indicating robust predictive capabilities.

Conclusions:

• The developed RFBW model offers an efficient and reliable method for classifying chronic venous insufficiency severity.
• Machine learning approaches, particularly with advanced optimization, show promise in improving the assessment of varicose veins.
• This predictive tool can aid in better management and treatment strategies for patients with venous disorders.