AUTOMATED DIAGNOSES FROM CLINICAL NARRATIVES - A MEDICAL SYSTEM BASEDON COMPUTERIZED MEDICAL RECORDS, NATURAL-LANGUAGE PROCESSING, AND NEURAL-NETWORK TECHNOLOGY

A collection of artificial, associative neural networks (PROMNET) inte rfaced to a computerized medical record is described Clinical narrativ es were subject to automated natural language processing, and relation s were established between 14,323 diagnoses and 31,381 patient finding s. Patient diagnoses and findings were counted, grouped into clinical entities, and used to train PROMNET. Training was completed in a few m inutes. PROMNET's dictionary contained about 20,000 words, and the neu ral network recognized more than 2800 disorders. Its performance was e valuated by an automated double-blind Turing test. PROMNET made clinic al decisions in a few seconds with sensitivity of 96.6%, and specifici ty of 95.7%. The most pertinent clinical entity was usually ranked hig hest. PROMNET is a powerful inference engine that learns from clinical narratives and interacts with medical personnel or patients in natura l language.