LAB PARSER: A PARSER FOR MEDICAL LAB DATA

摘要

A large volume of medical laboratory data enters the typical clinic. These files may be text or PDF format. We have designed and developed a prototype system, Lab Parser, to analyze files using natural language technology. The Parser uses recursive transition networks, sentence keyword scan to select appropriate rules, and rule macros for ease of coding. Benchmark files were employed for development. Test files yielded parsing rate average of 50%. After rule corrections, parsing score average increased to 75%. A module to transfer lab data to our electronic medical record has been developed and deployed. Storage of lab data into the record is fast and precise and data can then be incorporated into reports.