A method for analyzing acute exacerbation of chronic obstructive pulmonary disease

By constructing a directed acyclic graph and removing medication intervention variables, and combining variational autoencoders and graph convolutional networks, the problem of unclear causal relationships in the analysis of acute exacerbations of chronic obstructive pulmonary disease was solved, and stable early warning under different medication regimens was achieved.

CN122201733APending Publication Date: 2026-06-12PEOPLES HOSPITAL OF HENAN PROV
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
PEOPLES HOSPITAL OF HENAN PROV
Filing Date
2026-04-28
Publication Date
2026-06-12

AI Technical Summary

⚠Technical Problem

Existing methods for analyzing acute exacerbations of chronic obstructive pulmonary disease (COPD) lack robustness in early warning due to the failure to isolate spurious correlations caused by confounding factors such as medication and environment. These methods cannot maintain stability under different medication regimens or disease progression scenarios.

⚗Method used

A directed acyclic graph containing nodes of clinical symptoms, medication intervention, and environmental exposure is constructed. The causal strength is calculated through Granger causality test, medication intervention variables are removed, counterfactual feature distribution is generated, and pure causal features are extracted using variational autoencoders and graph convolutional networks. The warning level is output by combining Mahalanobis distance and risk classifier.

🎯Benefits of technology

It effectively eliminates the direct mapping interference of confounding factors, improves the stability and predictive performance of the analysis model under different medication regimens and disease scenarios, and ensures the accuracy and consistency of the early warning results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122201733A_ABST
    Figure CN122201733A_ABST
Patent Text Reader

Abstract

The present application relates to the technical field of medical data mining and auxiliary diagnosis, and discloses a chronic obstructive pulmonary disease acute exacerbation analysis method. Historical longitudinal data of a patient is acquired, a directed acyclic graph containing clinical symptom nodes, medication intervention nodes and environmental exposure nodes is constructed, and an edge corresponds to a causal strength parameter calculated based on a Granger causality test; current multidimensional monitoring features are received, and the specified medication intervention node features are subjected to out-edge truncation and feature zeroing intervention variable stripping processing according to the path coefficients in the directed acyclic graph, to generate counterfactual feature distribution. The present application excludes direct mapping interference of confounding factors such as medication intervention, removes false correlation between variables, overcomes the problem of prediction performance decay of correlation-based models in different medication scheme scenarios, and improves the stability of the analysis model when deployed across scenarios.
Need to check novelty before this filing date? Find Prior Art