Automated concept matching between laboratory databases

Abstract

To address the problem of semantic inconsistencies between medical databases, semantic network representations can be utilized to automate the matching of medical concepts between the databases. The performance of automated concept matching was tested by creating semantic network representations for two laboratory databases, one from a pediatric hospital and the other from an oncology institute. The matching algorithms identified all equivalent concepts that were present in both databases, and did not leave any equivalent concepts unmatched. By automatically identifying semantically equivalent concepts, the Medical Information Acquisition and Transmission Enabler (MEDIATE) facilitates data exchange between heterogeneous systems because no pre-negotiation is required. Consequently, system scalability and stability is improved.