Clinical cardiovascular risk interpretable prediction method based on causal inference

By constructing a structural causal model and combining it with cross-center transfer learning, the problem of insufficient causal reasoning in existing cardiovascular risk prediction methods is solved. Mechanism-level path decomposition and individualized interpretable prediction are achieved, improving the robustness and cross-center consistency of the model.

CN122455360APending Publication Date: 2026-07-24ANHUI HANXIN INTELLIGENT PROTECTION TECHNOLOGY CO LTD
0 Cites 0 Cited by

Patent Information

Application Number
CN202610731954.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-05-26
Publication Date
2026-07-24

AI Technical Summary

Technical Problem

Existing clinical cardiovascular risk prediction methods lack causal reasoning mechanisms and cannot reveal the path mechanisms and mediating relationships of disease occurrence. This results in poor model transferability, insufficient generalization ability, and insufficient interpretability, making it difficult to form cross-center consistent causal estimates and population-individual level prediction loops.

Method used

By integrating causal inference and cross-center transfer learning, a structural causal model is constructed to establish an identifiable formula for intervention-type mediation effects. Combining dual robust estimation and target maximum likelihood, robust and interpretable output middleware is generated, and cross-center consistent estimation is achieved through cross-domain transferable do-calculus.

Benefits of technology

It achieves mechanism-level pathway decomposition and individualized interpretable prediction of cardiovascular risk, and has strong robustness, high transferability and excellent clinical interpretability. It can obtain reliable prediction results even in the presence of exposure-induced confounding and time-varying covariates.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses a clinical cardiovascular risk interpretable prediction method based on causal inference, and the method comprises the following steps: collecting and normalizing multi-modal clinical data, and generating a causal diagram and a data baseline; establishing a recognizable formula of the intervention-type intermediary effect on the causal diagram; embedding the recognizable formula and the mechanism path set into a causal intervention type intermediary learning algorithm to obtain a robust estimation result and explainable output middleware; a cross-domain migratable do-calculus is introduced, a site is established, a node is selected, and migration correction is executed; and generating a crowd layer path decomposition result and an individual layer anti-fact contribution graph based on the cross-center consistent estimation set, and outputting an interpretable prediction result. According to the method, causal inference and cross-center transfer learning are fused, mechanism-level path decomposition and individualized interpretable prediction of cardiovascular risks are achieved, and the method has the advantages of being high in robustness, high in mobility and excellent in clinical interpretability.
Need to check novelty before this filing date? Find Prior Art