RISK: A random optimization interactive system based on kernel learning for predicting breast cancer disease progression

Abstract

Evaluating disease progression risk is a key issue in medicine that has been revolutionized by the advent of machine learning approaches and the wide availability of medical data in electronic form. It is time to provide physicians with near-to-the-clinical-practice and effective tools to spread this important technological innovation. In this paper, we describe RISK, a web service that implements a multiple kernel learning approach for predicting breast cancer disease progression. We report on the experience of the BIBIOFAR project where RISK Web Predictor has been developed and tested. Results of our system demonstrate that this kind of approaches can effectively support physicians in the evaluation of risk.