Semi-supervised self-learning driven medical text disease identification method

Abstract

Disclosed is a semi-supervised self-learning driven medical text disease identification method provided by the invention, feature classification of the medical text is mainly realized, the features include disease object types, disease progression, whether diseases occur, disease severity, disease condition, disease uncertainty and the like, and by identifying and classifying features, unstructured medical texts are structured and can be directly processed and used in further information mining; based on the features of few medical text annotation texts and many unannotated texts, the invention is developed from the aspects of feature extraction and classification model optimization; experimental results show that the method well makes up for the defects and deficiencies of few tagged texts, and since similar data and wrong classification data are easily introduced into the self-learning method, the effect of the semi-supervised SVM in the aspect of utilizing untagged data is better than that of the semi-supervised SVM, and the feasibility and high efficiency of the method are proved.

Description

technical field [0001] The invention relates to a medical text disease recognition method, in particular to a semi-supervised self-learning-driven medical text disease recognition method, which belongs to the technical field of medical text disease recognition. Background technique [0002] With the rapid development of information technology and the continuous improvement of the modern medical system, the informatization of medical records related to patients such as electronic medical records has been continuously enriched, and its importance has become increasingly prominent. It has gradually become an important guarantee for modern medical development and efficient management. Medical texts Electronic informatization is an inevitable trend of technological development and progress. With the rapid development of information technology, biomedical corpus is the product of changing the traditional writing and recording methods that are difficult to informatize and replacing...

AI Technical Summary

Problems solved by technology

[0006] First, due to insufficient software development and insufficient understanding of the existing medical document information management system, most medical information systems are far from reaching the standard and do not reflect the true value of these medical documents. There is a large amount of handwritten medical text information, the text information is seriously lacking in standardization, and the input system is not completely in accordance with the requirements, resulting in insufficient information availability. Various factors have caused the need to re-develop and utilize a large number of existing medical texts. In these precious texts The records imply important real information about the patient's condition, which not only helps the diagnosis and treatment of individual or similar patients, but also helps to understand the characteristics of this type of disease, and helps medical professionals to conduct research and actually prevent and treat diseases. Diagnosis and treatment, the development and utilization of existing technologies are insufficient;

[0007] Second, the natural language processing technology of the prior art has achieved good results in the general general field, and there are more common methods, but in the feature recognition of medical texts, due to the particularity and professionalism of medical texts Strong, the language structure is not the same as the general text corpus, coupled with the particularity of the feature relationship, the processing process relies more on professional dictionaries and knowledge, so it is more difficult in the actual feature identification process;

[0008] Third, the existing technology cannot realize the feature classification of medical texts, and cannot realize the identification and classification of the type of diseased object, the progress of the disease, whether the disease occurs, the severity of the disease, the condition of the disease, and the uncertainty characteristics of the disease. Existing technologies cannot make these unstructured medical texts structured and can be directly processed and used in further information mining; for the defects and deficiencies of medical texts with fewer labeled texts and more unlabeled texts, the self-learning method of the existing technology It is easy to introduce similar data and misclassified data, and the effect of using unlabeled data is not good, and the feasibility is poor;

[0009] Fourth, the existing self-learning text recognition method only considers a single optimal classification decision-making surface based on existing data, which may overfit the existing data, and the selection of model parameters has a certain degree of blindness, which makes the model deviate from rationalization , whether it is in the overall classification effect of multi-task or in the stricter record-based classification effect, the common Baseline model of the prior art is not ideal, and the accuracy and feasibility are not good.

Images