The Use of Computer-Assisted Case-Based Reasoning to Support Clinical Decision-Making - A Scoping Review

Clinical decision-making in healthcare is often difficult because of the increasing number of diagnostic findings and the overlap of disease characteristics. Information systems can be applied to support decision-making. Case-Based Reasoning (CBR) is used to solve a current case using similar past cases in the hospital's database. By assisting with diagnosis or treatment, CBR could improve the overall quality of healthcare for patients. However, the impact of CBR in clinical decision support has not yet been reviewed in depth in the literature. This scoping review highlights the properties of systems already in use, focusing on clinical applications, validation, interoperability, and case retrieval. A search query was performed in PubMed and Web of Science, and the results were selected for eligibility by title, abstract and full text screening. The following data items were observed: 'Publication year', 'Country of study', 'Disease group', 'Medical application', 'Patient number', 'Type of clinical data', 'Data interoperability', 'Type of similarity measure' and 'Expert validation'. The results showed that cancer and neoplasms are by far the most treated diseases, and demographic as well as historical data from patient or family records are frequently used as input. Most CBR systems were applied for therapy and diagnosis. More than 50% of all studies use data sets with > 100 records. About 24% of all systems were validated by experts, and 14% addressed data interoperability. CBR can be a useful approach to support clinical decision-making but needs further research to be used in clinical routine.