Medical text feature extraction method and system combined with NPL and large model, and medium

Abstract

The invention discloses a medical text feature extraction method and system combined with an NPL and a large model, and a medium, mainly relates to the technical field of text feature extraction, and is used for solving the problem that in-depth manual intervention is needed in an existing scheme, and key feature data in the existing scheme is manually processed and extracted. Comprising the following steps: acquiring desensitized diagnosis data for training; screening out a preset number of training medical record documents with the highest association degree, and obtaining a trained BERT model; taking an output layer of the trained BERT model as an input layer of an NPL algorithm; processing data annotation of the medical record document through the trained BERT model; the association relationship between the annotation data and the symptom keyword and the diagnosis disease name is used as the input of an NPL algorithm, the relationship between entities is output, and then structured data is generated; and generating a running script for extracting the relationship between the entities from the structured data by using the large generative model.

Images