Firefly-SVM predictive model for breast cancer subgroup classification with clinicopathological parameters

Area of Science:

• Oncology
• Bioinformatics
• Machine Learning

Background:

• Breast cancer is a prevalent disease in women with diverse biological and clinical characteristics.
• Machine learning (ML) offers potential for analyzing complex medical data in breast cancer detection and prognosis.

Purpose of the Study:

• To develop a data-driven breast cancer model for enhanced subtype classification.
• To accurately distinguish between triple-negative breast cancer (TNBC) and non-triple-negative breast cancer (non-TNBC).

Main Methods:

• Proposed a firefly-support vector machine (SVM) model for breast cancer prediction.
• Utilized clinicopathological and demographic data from tertiary cancer centers.
• Employed hyperparameter tuning for optimizing the SVM model.

Main Results:

• The firefly-SVM model achieved superior prediction accuracy (93.4%) compared to Grid-SVM, PSO-SVM, and GA-SVM.
• Model performance was validated using metrics like F1-score, MSE, AUC, log loss, and PR curves.

Conclusions:

• The firefly-SVM model serves as a valuable tool for breast cancer subgroup classification.
• This approach can aid clinicians in patient management and treatment decisions.