A concept extraction method for Chinese electronic medical records based on deep learning

Abstract

The invention discloses a method for extracting concepts in a Chinese electronic medical record based on deep learning. The method comprises the followings steps of: training context distributed characteristics of target words by utilizing a deep architecture of a multi-layer sparse autocoder, then, combining the context distributed characteristics with entity characteristics having advanced concept meanings, namely, marking characteristics and part of speech to form whole characteristics, inputting the whole characteristics in a deep belief network to perform model training, comparing the marking characteristics of samples to perform residual calculation, and optimally adjusting the performance of the whole deep architecture through supervised fine-adjustment. According to the method disclosed by the invention, the feature that the characteristics are deeply optimized by the deep learning is sufficiently utilized; simultaneously, the entity characteristics are added and used as prior knowledge; and thus, the classification and prediction accuracy can be increased while dependence on artificial characteristics is reduced.

Description

technical field [0001] The invention relates to a method for extracting concepts from Chinese electronic medical records based on deep learning. Background technique [0002] Electronic Medical Record (EMR) refers to digital information such as words, symbols, charts, graphics, data, images, etc. generated by medical personnel using medical information systems during medical activities, and the activity records can be transmitted and reproduced , and use information technology to store and manage. With the continuous popularization of electronic medical records, a large amount of medical data has been continuously accumulated in the form of electronic medical records. Among them, a considerable part of the data still exists in the form of narrative texts. How to extract medical concepts in electronic medical records and structure unstructured texts has become an urgent problem to be solved in the development of electronic medical records. [0003] Concept extraction refers...

AI Technical Summary

Problems solved by technology

[0006] Aiming at the problems that the traditional concept extraction method relies too much on manually formulating features, the existing concept extraction method based on deep learning lacks important prior knowledge, and the accuracy rate has not made breakthrough progress, a Chinese electronic technology based on deep learning is proposed. A medical record concept extraction method that combines deep learning methods with a small number of entity features

Images