Causal Reasoning Methods in Medical Domain: A Review

Causal reasoning has been a key topic in medical domain with many applications, in which the core problem is to infer the causal effects of medical treatments with data mining However, there are obstacles such as unstable identification and false associations when applying traditional machine learning methods dealing with the effect estimation about medical treatments due to the large-scale and high-dimensionality of medical data. Furthermore, there is no thorough survey of causal reasoning methods for medical domain problems, which is an emerging research direction. To meet the challenge, the causal reasoning in medical domain is surveyed to systematically classify and summarize causal reasoning methods in two dimensions: four categories of core ideas and three levels of causal structure. The thorough review demonstrates that causal reasoning methods have theoretical and practical significance in medical domain, which is a research field full of potential.