In a narrative review published in Infectious Agents and Cancer, researchers evaluated how artificial intelligence can be integrated into Mendelian randomization workflows to strengthen causal inference across genomics, epidemiology, and oncology. Mendelian randomization uses genetic variants as instrumental variables to assess whether observational associations between exposures and health outcomes reflect true causal relationships. The review examines how recent developments in machine learning, deep learning, and large language models can augment this analytical process across large-scale biomedical datasets.
The review was structured according to the Scale for the Assessment of Narrative Review Articles, known as SANRA. To synthesize methodological developments, practical implementations, and emerging analytical frameworks, the authors conducted comprehensive literature searches across PubMed, Scopus, and the Web of Science. The resulting synthesis characterizes how computational intelligence tools intersect with epidemiological causal modeling across distinct phases of research execution.
The authors identified three operational tiers where artificial intelligence interfaces with the Mendelian randomization pipeline: before, within, and around the core analysis. Upstream of the causal calculation, artificial intelligence algorithms assist in phenotype construction, exposure discovery, and biological target prioritization.
Within the analytical phase, artificial intelligence supports the prioritization of biologically informed instruments and facilitates nonlinear causal estimation. However, the authors emphasized that these algorithmic tools remain entirely dependent on standard instrumental-variable assumptions.
While these innovations offer enhanced efficiency, scalability, and high-dimensional data processing, the authors concluded that algorithmic power cannot substitute for biological plausibility, instrument validity, transparent methodology, or human oversight.
The stated objective of the review was to critically examine the emerging integration of artificial intelligence and Mendelian randomization, highlighting current applications, methodological opportunities, limitations, and future directions. These records were selected to illustrate methodological developments, practical applications, and emerging computational frameworks capable of supporting causal inference across large-scale and high-dimensional biomedical datasets.
The authors emphasized that artificial intelligence operates across distinct stages without altering the core logic of causal deduction. When deployed around the analytical pipeline, tools facilitate workflow automation, data retrieval, reporting, and evidence synthesis without modifying causal inference itself. Causal inference continues to depend strictly on biological plausibility, instrument validity, and transparent analytical design.
In oncology, these emerging frameworks offer clear opportunities for biomarker discovery and therapeutic target prioritization. Promising approaches, including Deep Mendelian Randomization, foundation models, and analytical agents based on large language models, provide improved efficiency, scalability, and data integration. However, the review concluded that these innovations require rigorous methodological validation. Future progress will rely on transparent, explainable, and human-supervised artificial intelligence frameworks that strictly adhere to established Mendelian randomization principles and robust methodological design.
Subject of Research: Artificial intelligence integration in Mendelian randomization for oncology causal inference
Article Title: Artificial intelligence-assisted Mendelian randomization: a methodological framework for causal inference with applications in oncology
Article References: Cascella, M., Granata, V., Fusco, R., Simonetti, I., Porto, A., Di Monaco, P., Natangelo, S., Zarrella, A., Esposito, D., Pagano, A., De Feo, R., Donnarumma, D., & Santoriello, V. (2026). Artificial intelligence-assisted Mendelian randomization: a methodological framework for causal inference with applications in oncology. Infectious Agents and Cancer. https://doi.org/10.1186/s13027-026-00799-8
Image Credits: AI Generated
DOI: 10.1186/s13027-026-00799-8
Keywords: Mendelian randomization, artificial intelligence, causal inference, oncology, machine learning, deep learning, large language models, DeepMR, biomarker discovery, phenotype construction, exposure discovery, target prioritization, instrumental variables, narrative review, SANRA guidelines, genomics
News Source: Blake Davidson. (November 19, 2026). Artificial Intelligence Enhances Mendelian Randomization Frameworks for Oncology Causal Inference. Scienmag.



