Ranking-based Artificial Intelligence: from clinical decision support to precision medicine.
Due to the rapid growth of information available about individual patients, most physicians suffer from information overload and inefficiencies when they review patient information in health information technology systems. To address this issue, we developed ranking-based AI/ML systems to improve information retrieval from electronic health records (EHRs) and to facilitate clinical decision support. Our systems leverage the key ideas from Recommender Systems for prioritizing patient information based on various similarities among physicians, patients and information items. The need for information prioritization/selection is not unique only in EHR systems; for example, selecting the right drugs for the right patients is a primary goal of precision medicine. With similar ranking-based AI/ML techniques, we also tackled the problem of cancer drug selection in a learning-to-rank framework, and formulated the cancer drug selection problem as to accurately predicting the ranking positions of sensitive drugs for each patient. In this talk, I will be presenting our ranking-based AI/ML methods for clinical decision support and for precision medicine, and showcase the effectiveness of ranking-based AI/ML techniques for information prioritization in general application scenarios.