An ultrasonic report intelligent quality control method and system fusing a rule engine and a large language model

Abstract

An ultrasonic report intelligent quality control method fusing a rule engine and a large language model, comprising: data acquisition and analysis: by a computing node, real-time access to a target ultrasonic report to be issued in front of an ultrasonic workstation, and analysis of the target ultrasonic report to obtain a normalized report, the normalized report being represented in an object tree form in a memory buffer of the computing node, the normalized report including at least a patient attribute field, an examination type field, a structured measurement item field, an unstructured text description field and a report full life cycle timestamp field. The present application realizes the paradigm expansion from 'after-the-fact inspection' to 'intervention in the process', performs quality control at the moment before the ultrasonic report is issued and issues intervention instructions in real time, and completes interception or reminder before the unqualified report flows into the clinical process, thereby guarding the medical quality bottom line from the source.