Clinical decision support method and system for intelligent disease screening

Abstract

The invention provides a clinical decision support method and system for intelligent disease screening. A disease screening model can improve disease screening accuracy and reduce misclassification cost. The method comprises: performing post-structuring on an inspection report to obtain a sample data set; training an individual classification model according to the sample data set to obtain a plurality of homogeneous classification models; setting the classification threshold of an integrated model at least by using a minimum misclassification cost threshold selection method MCTSM to obtain adisease screening model, wherein the input of the disease screening model is the inspection report, the output of the disease screening model is a disease classification result, and the disease classification result is used for supporting clinical decision.

Description

technical field [0001] The invention relates to the field of medical technology, in particular to a clinical decision support method and system for intelligent disease screening. Background technique [0002] Traditionally, the diagnosis and screening of cancer (such as gastric cancer, breast cancer, etc.) rely on the analysis of medical records and examination reports by doctors. However, due to the increasingly heavy work pressure, lengthy medical records and examination reports, which have a greater impact on the work efficiency of doctors, as well as the difficulty of cancer diagnosis and screening work itself and the limitations of the professional level of grassroots doctors, leading to cancer screening. There is a high rate of misdiagnosis and missed diagnosis. [0003] With the advancement of artificial intelligence technology in recent years, data-driven analytical research has gradually become a strong support and supplement for cancer-related research in the clin...

AI Technical Summary

Problems solved by technology

Another example is to incorporate cancer comorbidities into cancer diagnosis and treatment research, and consider using accuracy, sensitivity, and specificity as evaluation indicators for model performance, but ignores the accuracy rate caused by the imbalance of cancer-related data sets, etc. Situations where the use of indicators is limited

[0004] Usually, the classification model established by machine learning method inevitably has generalization error, and the selection of classification threshold is different, which will bring different error losses such as sensitivity and specificity.

This error loss is manifested as misdiagnosis and missed diagnosis in gastric cancer screening, that is, wrongly judging a cancer-free sample as having cancer and wrongly judging a cancer-bearing sample as non-cancerous. The former will cause further inspection difficulties, and the latter may lead to medical malpractice

[0005] Clinically, the cost losses caused by misdiagnosis and missed clinics are not equal. However, how to reduce the cost of misclassification and improve the accuracy of disease screening has not been further considered by those skilled in the art

Images