Medical text named entity identification method based on pre-training model and fine tuning technology

Abstract

The invention provides a medical text named entity recognition method based on a pre-training model and fine tuning technology, and the method comprises the steps: firstly carrying out the pre-training of a BERT pre-training model through a large-scale unstructured electronic medical record and other medical texts, so as to train a pre-training model containing semantic representation informationin a text; and carrying out fine tuning on the generated pre-trained model by utilizing the stacked extended convolutional neural network so as to obtain a deep neural network model capable of carrying out automatic identification on the named entities in the medical field. According to the pre-training model provided by the invention, semantic information in the text can be captured more accurately, the semantic information can be migrated to a specific task more effectively, and the named entity recognition accuracy of the model is improved. The stacking expansion convolutional neural network and the pre-training model are combined to finely adjust the model, and finally, the named entities of the medical text are identified, so that semantic information in the text can be well captured,and parallel computing can be performed to improve the model training speed.

Description

technical field [0001] The invention relates to the technical field of data mining, in particular to a medical text named entity recognition method based on a pre-training model and fine-tuning technology. Background technique [0002] Clinical medicine is a science that studies the etiology, diagnosis, treatment and prognosis of diseases, improves the level of clinical treatment, and promotes human health. Medical text data such as electronic medical records in clinical medicine has great value for research in the medical field. With the popularity of the Internet, more and more online clinical medical communities and clinical medical consultation websites have emerged. These online clinical medical websites also generate rich medical text data. There are a large number of real personal cases in these medical texts, which have a wealth of clinical medical value hidden. But most of these clinical medical texts are in an unstructured state. In order to fully tap the value...

AI Technical Summary

Problems solved by technology

Compared with the previous two methods, this method does not require strong linguistic knowledge, has high flexibility and adaptability, but the effect of named entity recognition depends on the quantity and quality of labeled data

Images