Practice-based Evidence in Medicine: Where Information Retrieval Meets Data Mining

A new approach in medical practice is emerging thanks to the increasing availability of large-scale clinical data in electronic form. In practice-based evidence [5, 6], the clinical record is mined to identify patterns of health characteristics, such as diseases that co-occur, side-effects of treatments, or more subtle combinations of patient attributes that might explain a particular health outcome. This approach contrasts with what has been the standard of care in medicine, evidence-based practice, in which treatment decisions are based on (quantitative) evidence derived from targeted research studies, specifically, randomised controlled trials. Advantages of consulting the clinical record for evidence rather than relying solely on structured research include avoiding the selection bias of the inclusion criteria for a clinical trial and monitoring of longer-term outcomes and effects [5]. The two approaches are, of course, complementary — a hypothesis derived from large-scale data mining could in turn form the starting point for the design of a clinical trial to rigorously investigate that hypothesis. Information retrieval can play an important role in both approaches to collecting medical evidence. However, the use of information retrieval methods in collecting practice-based evidence requires moving away from traditional documentoriented retrieval as the end goal in itself, to viewing that retrieval as an intermediate step towards knowledge discovery and population-scale data mining. Furthermore, it may require the development of more context-specific retrieval strategies, designed to identify specific characteristics of interest and support particular tasks in the medical context.