Clinical automatic diagnosis method and system based on hierarchical structure and co-occurrence structure

Abstract

The invention discloses a clinical automatic diagnosis method and system based on a hierarchical structure and a co-occurrence structure, and the method comprises the steps: collecting basic data, andcarrying out the preprocessing of the basic data, and the basic data at least comprises an electronic health record of a patient, a diagnosis code set of the patient, and a hierarchical structure between diagnosis codes; modeling clinical record text features based on the preprocessed basic data; modeling diagnosis code semantic representation based on the preprocessed basic data; and based on the clinical record text feature modeling result and the diagnosis code semantic representation modeling result, performing hierarchical allocation modeling on the diagnosis code to obtain a clinical automatic diagnosis result. According to the invention, by introducing modeling of the hierarchical structure, high-level diagnosis results can be used for guiding distribution of low-level diagnosis codes, so that the problem of unbalanced distribution is solved, and meanwhile, the accuracy of clinical automatic diagnosis can be further improved by capturing correlation between the diagnosis codesthrough a co-occurrence structure.

Description

technical field [0001] The invention relates to the technical field of medical data mining, in particular to an automatic clinical diagnosis method and system based on a hierarchical structure and a co-occurrence structure. Background technique [0002] Clinical records are an important part of electronic health records (EHRs), containing lengthy and terminological text descriptions such as medical history, chief complaints, current symptoms, and laboratory test results. To avoid duplication and ambiguity caused by large amounts of text, the World Health Organization recommends using diagnostic codes from the International Classification of Diseases (ICD) to represent each disease, symptoms and signs, etc. The goal of clinical diagnosis is to assign the most probable diagnosis code to a patient based on clinical records. Traditional clinical diagnosis is done by trained clinical coders, and manual diagnosis remains labor-intensive and error-prone due to the large and evolvi...

AI Technical Summary

Problems solved by technology

[0010] In view of this, the present invention provides a clinical automatic diagnosis method based on hierarchical structure and co-occurrence structure. By introducing modeling of hierarchical structure, high-level diagnostic results can be used to guide the allocation of low-level diagnostic codes, thereby Solving the problem of uneven distribution, and capturing the correlation between diagnostic codes through co-occurrence structure can further improve the accuracy of clinical automatic diagnosis

Images