Utilizing Directed Acyclic Graphs in the Study of Digestive Cancers
10.52927/jdcr.2025.13.3.251
- Author:
Myeongjee LEE
1
Author Information
1. Biostatistics Collaboration Unit, Department of Biomedical Systems Informatics, Yonsei University College of Medicine, Seoul, Korea
- Publication Type:REVIEW ARTICLE
- From:
Journal of Digestive Cancer Research
2025;13(3):251-259
- CountryRepublic of Korea
- Language:English
-
Abstract:
Directed acyclic graphs (DAGs) have emerged as a foundational framework for causal inference in medical and epidemiological research, offering a transparent and systematic approach to visualizing causal assumptions and informing analytic choices. By mapping hypothesized relationships among exposure, outcome, and other covariates, DAGs help researchers identify confounding, mediation, and collider structures, determine appropriate adjustment sets, and avoid bias from inappropriate variable control. In digestive cancer research, DAGs have been especially useful for clarifying complex causal pathways, improving the validity of observational analyses, and informing advanced study designs such as mediation analysis and Mendelian randomization. This paper outlines the conceptual foundations of DAGs, including d-separation, the backdoor criterion, and M-bias, and demonstrates their practical application through the web-based tool DAGitty. We also describe how published digestive cancer studies have incorporated DAG-based frameworks to strengthen study design, ensure appropriate variable adjustment, and improve the transparency and reproducibility of causal interpretation. Collectively, these approaches show that DAGs are not merely graphical aids but essential methodological tools for causal reasoning in contemporary biomedical research.
