A case-based ensemble learning system for explainable breast cancer recurrence prediction.

Significant progress has been achieved in recent years in the application of artificial intelligence (AI) for medical decision support. However, many AI-based systems often only provide a final prediction to the doctor without an explanation of its underlying decision-making process. In scenarios concerning deadly diseases, such as breast cancer, a doctor adopting an auxiliary prediction is taking big risks, as a bad decision can have very harmful consequences for the patient. We propose an auxiliary decision support system that combines ensemble learning with case-based reasoning to help doctors improve the accuracy of breast cancer recurrence prediction. The system provides a case-based interpretation of its prediction, which is easier for doctors to understand, helping them assess the reliability of the system's prediction and make their decisions accordingly. Our application and evaluation in a case study focusing on breast cancer recurrence prediction shows that the proposed system not only provides reasonably accurate predictions but is also well-received by oncologists.