A Care Oriented Decision Support System Based on Ensemble Methods

Predicting the response to drug treatments is a critical need in the field of precision medicine and oncology. Indeed, one of the limitations of wide spectrum treatments is that predictive response factors are still poorly known: among such factors, the elements predicting in advance if a patient can sustain the treatment. The advantage of having such a prediction can influence both the patient care process, and the management of healthcare structures with cost optimisations. In this chapter, a Decision Support System (DSS) able to estimate the capability of patients to sustain oncologic therapy is described. Such a framework is based on Machine Learning (ML) techniques and on the usage of ensemble methods, to maximise the prediction reliability. The ML algorithms can build sub-learners of the ensemble model on the base of both general information about the patients and temporal data related to the oncological therapies of these patients, with the information of success/failure of the therapy. The proposed DSS is realised within the framework Oncological Advanced Therapy Management (OATM) project, whose aim is to improve the quality of care for oncology patients through solutions that optimise the process.