Causal inference method and system for cascaded medical observation data

Abstract

The invention discloses a causal inference method and system for cascaded medical observation data, and the method comprises the steps: building an improved cascaded nonlinear additive noise model with reasons in a causal relationship, an intermediate variable corresponding to each depth in a cascaded structure, and a result in the causal relationship as parameters, medical observation data with a cascade structure can be better matched, the accuracy of identifying the causal direction of the cascade medical data is improved, meanwhile, a variational lower bound corresponding to a maximized edge log-likelihood function is solved through a preset adversarial training model, KL divergence is bypassed by utilizing an adversarial strategy instead of an approximation formula, additive noise can be allowed to have wider distribution, so that the inference capability of the model is improved, and compared with the prior art, the causal direction of the medical observation data with the cascade structure can be well identified, and the causal direction identification accuracy is remarkably improved.

Description

technical field [0001] The invention relates to the technical field of medical data analysis, in particular to a causal inference method and system for cascading medical observation data. Background technique [0002] With the advent of the era of big data, a large amount of data has been generated in various fields, and it is very important to study the causal relationship between these data. Causal inference is widely used in biomedicine. Biologists use the observed disease gene data to study the causal relationship between a certain disease and genes; starting from the comprehensive information of medicine and biology of drugs, infer the cause of adverse drug reactions. Molecular factors; exploiting genetic data to discover causal molecular interactions. In addition, causal inference is also widely used in other fields, such as using causal networks to predict economic models; using causal graph models to study the performance of TCP network protocols, etc. [0003] At ...

AI Technical Summary

Problems solved by technology

However, the existing causal inference methods do not take into account that in real data, there may not be a direct causal influence between the causal variable and the outcome variable, there may be intermediate variables between the cause and the outcome, and there may be an indirect non-causal relationship between the initial cause and the final outcome. Linear causal effects, so existing causal inference methods do not perform satisfactorily on data with a cascaded structure

In addition, although causal inference has achieved a lot of results in medical treatment, there is currently no way to study this indirect, cascading medical data from observational data.

[0004] In view of this, how to provide a way to start from observation data, infer the causal direction of indirect medical observation data with cascade structure, improve the accuracy of causal direction identification, and solve the problem of medical data with cascade structure not considered in the existing methods The causal inference method of the cascade medical observation data is a technical problem to be solved by those in the technical field

Images