Medical relationship extraction method and device

Abstract

The invention discloses a medical relationship extraction method and device, and the method comprises the following steps: carrying out the statistics of the occurrence frequency of a medical concept pair in a set time window from a medical electronic medical record, and obtaining two medical concept vectors in the medical concept pair; matching the two medical concepts with a knowledge base to obtain an incidence relation between the two medical concepts so as to construct a relation concept triple; mining a plurality of concept statements from a medical text set according to the relation concept triple; constructing a training sample set, wherein the training sample set comprises positive samples and negative samples, and each sample structure is composed of a relation concept triple, two medical concept vectors and a concept statement; training a fusion model by using the training sample set to obtain a trained fusion model; and performing medical relationship extraction by using the trained fusion model. According to the method, the relationship between the medical concepts can be continuously mined. Chapter titles are introduced to form concept statements, and effective training samples are increased.

Description

technical field [0001] The invention relates to the field of information technology, in particular to a method and device for extracting medical relations considering medical texts and medical electronic medical records. Background technique [0002] In recent years, many models for relation mining have emerged, including pattern matching and machine learning. With the rapid development of deep learning, researchers began to use neural network models in relation extraction, using word embedding and position embedding vectors as input, and using recurrent neural networks. The sentence-level attention mechanism originally used in machine translation is now also applied to relation extraction to automatically capture important words and sentences, becoming a necessary mechanism in the model. However, these supervised models lack real medical relational data for training. Due to the high cost of manual labeling, neural networks do not have enough sample sentences, so distant s...

AI Technical Summary

Problems solved by technology

If the method of rule matching is adopted, the coverage and accuracy of the acquired relationship and medical knowledge are obviously insufficient

And if you want to use machine learning or deep learning for automatic relationship extraction, there is no existing training set and model for target relationship mining, so it is impossible to quickly and parallelly mine the target relationship

Therefore, there can be a plan to do similar things, but the accuracy, timeliness and extensiveness of the invention cannot be achieved

Images