Integrating knowledge from Case Report: a medical-ontology based multimodal information system with structured summary

Abstract

Abstract Published Case Report literature, as an important medical information carrier, records invaluable new discoveries of rare diseases, diagnoses and novel therapies as well as unique clinical experiences. Millions of Case Reports following formatted content have been collected in PubMed sharing medical knowledge across worldwide healthcare community. However, the ability of searching and viewing a vast number of Case Reports, and grasping most relevant and essential content in a time-constrained environment has been severely limited with most existing keyword-based retrieval tools due to their inherent limitation in handling the scale, unstructuredness and heterogeneity of Case Reports. To alleviate such limitations, we have developed CRFinder, a comprehensive multimodal information system with powerful interfaces for searching, visualizing and analyzing Case Report. In this paper, we will discuss the extraction pipeline of clinical knowledge in Case Report and core functional components of proposed system. The efficacy of our pipeline and system is demonstrated using 52949 open access case reports published from 2000 to 2021 in PubMed Central, from which CRFinder builds structured case summary by automatically extracting clinical and biological entities, medical images of different modalities. All the different types of essential information can be conveniently and effectively searched, displayed and analyzed in a well-structured and correlated manner. The access of CRFinder system is now available online (http://49.235.104.183:8080/).