SURVIVALSENSE: SVM-DRIVEN UTILITY KERNEL FOR BREAST CANCER OUTCOME ESTIMATION

Abstract

Accurate prediction of breast cancer survival is critical for effective treatment planning and patient care. Traditional statistical models often struggle to handle complex, high-dimensional clinical and genomic data, limiting their predictive accuracy. This study presents SurvivalSense, a novel framework utilizing a Support Vector Machine (SVM) utility kernel to estimate breast cancer survival outcomes. By integrating clinical features, histopathological data, and molecular markers, the utility kernel enhances the SVM’s ability to capture nonlinear relationships and prioritize critical prognostic factors. Experimental evaluation on benchmark datasets demonstrates that SurvivalSense achieves superior predictive accuracy, robustness, and interpretability compared to conventional SVMs and other machine learning models. The results highlight the potential of SVM utility kernels to improve survival estimation, support personalized treatment strategies, and facilitate better clinical decision-making in oncology