Artificial Intelligence in cancer epigenomics: a review on advances in pan-cancer detection and precision medicine

Abstract

DNA methylation is a fundamental epigenetic modification that regulates gene expression and maintains genomic stability. Consequently, DNA methylation remains a key biomarker in cancer research, playing a vital role in diagnosis, prognosis, and tailored treatment strategies. Aberrant methylation patterns enable early cancer detection and therapeutic stratification; however, their complex patterns necessitates advanced analytical tools. Recent advances in artificial intelligence (AI) and machine learning (ML), including deep learning networks and graph-based models, have revolutionized cancer epigenomics by enabling rapid, high-resolution analysis of DNA methylation profiles. Moreover, these technologies are accelerating the development of Multi-Cancer Early Detection (MCED) tests, such as GRAIL's Galleri and CancerSEEK, which improve diagnostic accuracy across diverse cancer types. In this review, we explore the synergy between AI and DNA methylation profiling to advance precision oncology. We first examine the role of DNA methylation as a biomarker in cancer, followed by an overview of DNA profiling technologies. We then assess how AI-driven approaches transform clinical practice by enabling early detection and accurate classification. Despite their promise, challenges remain, including limited sensitivity for early-stage cancers, the black-box nature of many AI algorithms, and the need for validation across diverse populations to ensure equitable implementation. Future directions include integrating multi-omics data, developing explainable AI frameworks, and addressing ethical concerns, such as data privacy and algorithmic bias. By overcoming these gaps, AI-powered epigenetic diagnostics can enable earlier detection, more effective treatments, and improved patient outcomes, globally. In summary, this review synthesizes current advancements in the field and envisions a future where AI and epigenomics converge to redefine cancer diagnostics and therapy.

Keywords: Artificial intelligence; Cancer epigenomics; DNA methylation; Deep learning; Early detection; Liquid biopsy; Machine learning; Multi-cancer diagnostics; Multi-omics; Pan-cancer; Precision oncology.

Fig. 1

Schematic flow diagram of the review design. The schema depicts five main phases of the review process, including DNA methylation and its significance, Profiling methods, AI-driven cancer detection, MCED pipelines and technologies, and Challenges and future directions

Fig. 2

Mechanism of aberrant DNA methylation and its impact on Cancer Cell proliferation: In healthy cells, promoter hypomethylation activates tumor suppressor genes, while hypermethylation inactivates oncogenes. Conversely, in cancerous cells, hypermethylation silences tumor-suppressing genes, and hypomethylation activates cancer-promoting genes. These epigenetic alterations contribute to cancer-related processes and can be accessed for early detection, prognosis, biomarker identification, understanding tumor microenvironment dynamics, and assessing disease progression

Fig. 3

Illustration of the machine learning (ML) lifecycle: The figure depicts key stages of ML lifecycle, represented as interconnected gears to emphasize the iterative nature of the process. It represents the continuous cycle of model development training, assessment, and implementation, illustrating the transition from one phase to the next

Fig. 4

AI-driven Framework for detecting and classification multiple cancer signals: A visual depiction of the AI-based system designed to identify and classify signals from multiple cancers, highlighting the essential components of the machine and deep learning algorithms implemented in these processes

Fig. 5

Timeline depicting advancements in Multi-Cancer Early Detection (MCED) from 2021 to 2025, highlighting progress in liquid biopsy, multi-omics integration, validation studies, and clinical implementation for early cancer screening