Total pulmonary respiration medicine disease medical auxiliary diagnosis system based on multi-step decision

Abstract

The invention discloses a total pulmonary respiration medicine disease medical auxiliary diagnosis system based on multi-step decision, and the system comprises the following steps: S1, judging whether the diseases are respiratory medicine department diseases or not, and entering a next stage when any keyword appears in an input description; s2, checking blood routine fields, and entering the next stage when the number of eosinophils in the blood routine is not increased; using the prediction rejection module, and entering the next stage when the classification result of the classifier is that the disease can be classified; and S4, using a disease diagnosis classification module. The system has the beneficial effects that (1) the overall generalization ability is obviously improved; 2) a multi-step decision-making method is adopted, different field information is used in different decision-making steps, and data can be reasonably utilized to the greatest extent, 3) a self-attention mechanism is used in the method, 4) a large-scale pre-training language model ALBERT is used in the method, the generalization performance is obviously improved, and the overall accuracy is improved by about 10%.

Description

technical field [0001] The invention belongs to the technical field of medical aided diagnosis, and in particular relates to a medical aided diagnosis system for whole lung and respiratory diseases based on multi-step decision-making. Background technique [0002] Electronic medical records (EMR) contain a wealth of patient clinical data information, such as history of present illness, clinical manifestations, imaging reports, etc. With the development of artificial intelligence deep learning technology, how to use rich and large-scale electronic medical record data to build a medical auxiliary diagnosis system and provide artificial intelligence technical support for doctors has become an urgent problem to be solved. [0003] However, there are still many deficiencies and defects in the currently known artificial intelligence systems constructed by combining deep learning and electronic medical records. [0004] First, sufficient data cleaning and feature engineering have ...

AI Technical Summary

Problems solved by technology

Electronic medical records (EMR) are generally unstructured data. Existing deep learning models can fit the existing data well, but the model has poor generalization ability and serious overfitting, making it difficult for the model to be used in practice

Images