Demetrios Prize 2019: Machine Learning as an Adjunct to Medical Decision Making

One of the winners, Nicolas Woitzik from Germany, evaluated in his thesis the textual clinical recommender system which represents a computerized clinical decision support system (CDSS). It analyses the patient’s discharge summaries with the help of information retrieval and natural language processing methods. It provides the user with a similar patient case out of a database to include this information into the user's decision-making process. We conducted an experiment to validate the correlation between the computed similarities by the new CDSS and the similarity judgment of medical experts, junior doctors, and medical students. Taken together, the retrieval system still needs improvements, either based on an improved retrieval algorithm or by additional features. However, it is likely that the performance of the system will improve the more discharge summaries a database contains like it was shown in this thesis. Our data suggest that the simrec software might indeed become an important clinical tool to share clinical experience between hematologists and possibly also other medical specialties.