Standardization of clinical terminology based on hybrid recall and Ernie

Abstract

The task of clinical terminology standardization is an indispensable part of medical information statistics works. Clinical terms often have problems with colloquial terminology and diversity of expression in clinical practice. Therefore, there are many different ways of writing the same diagnosis, procedure, symptom, etc. Clinical terminology standardization is to correspond a descriptive text of a non-standard Chinese clinical term to a given standard word in the Chinese clinical terminology database. This paper focuses on Chinese clinical texts, ERNIE-health is used as an implied semantic scoring module to generate candidate answers based on text similarity retrieval, and tested on the Chinese data set of medical professionals based on SNOMED CT annotation for the first time. Experiments show that this method has achieved good results in practical application.