Information retrieval system for medical narrative reports

This paper presents a novel information retrieval system designed specifically for medical case finding applications. The proposedsystem begins by extracting medical information from free-text narrative reports and storing it in a predefined relational clinical data mart. Theextraction is performed using a medical thesaurus and a regular expression pattern match. Following the extraction phase, inclusion/exclusioncriteria are provided to the system using a physician-friendly user interface. The system converts the entered criteria into a single SQL commandwhich can be then executed on the relational data mart. In order to achieve the appropriate response time required for on-line analysis, thesystem implements several caching mechanisms. The proposed system has been examined on real-world database. The performance of the systemhas been compared to the results obtained manually by a physician. The comparison indicates that the proposed system can be used for noncriticalcase-finding applications such as: finding appropriate patients for clinical trials.