Directed Acyclic Graphs in Oncology Research: Applications and Illustrated Example

Area of Science:

• Oncology
• Epidemiology
• Causal Inference

Background:

• Cancer development involves multifactorial risk factors, necessitating clear understanding of causal relationships.
• Observational studies in oncology require robust methods to identify and address confounding, mediation, and colliders.

Purpose of the Study:

• To present the development and application of Directed Acyclic Graphs (DAGs) in a real-world skin cancer observational study.
• To provide a practical guide for constructing and utilizing DAGs to control for confounding in cancer research.

Main Methods:

• Development and application of a Directed Acyclic Graph (DAG) framework.
• Utilizing DAGs for variable selection to manage confounding in observational data.
• Illustrative case study involving skin cancer research.

Main Results:

• Demonstrated the practical utility of DAGs in a real-world skin cancer study.
• Outlined actionable steps for DAG construction and confounder identification.
• Enhanced the ability to estimate causal effects in observational oncology research.

Conclusions:

• DAGs are essential tools for elucidating complex causal relationships in cancer research.
• This approach improves the transparency and validity of causal claims in oncology.
• Provides a valuable guide for researchers across various cancer research domains.