Artificial Intelligence in Cancer Epigenetics: A Review of How Machine Learning Is Revealing New Patterns Beyond DNA Mutations
Fortune Itoje Ebiala
*
Department of Microbiology, Faculty of Life Sciences, University of Benin, Benin.
Cecilia Chika Unegbu
Department of Natural Sciences, Bowie State University, 14000 Jericho Park Rd, Bowie, MD 20715, United States.
Precious Oluwamosope OKUNOLA
Public Health In-vitro Diagnostics Control Laboratory, Medical Laboratory Science Council of Nigeria, Yaba, Lagos State, Nigeria.
Ogochi Blessing Chukwuneke
Department of Biotechnology, Middle Tennessee State University, USA.
Ololade Funke Olaitan
David Eccles School of Business, information Systems, University of Utah, Zip Code 84112, United States.
Ndubisi Ubaka Edebeatu
Texila American University, Guyana.
Damilola Ayodele Ojo
College of Business, Missouri State University, Springfield, Missouri, USA.
*Author to whom correspondence should be addressed.
Abstract
Beyond the human genetic codes, epigenetics also plays a key role in regulating how genes are expressed. While mutations or genes are known to be major causes of cancer, Epigenetic alterations, including DNA methylation, histone modifications, and non-coding RNAs, play critical roles in cancer progression by altering gene expression without involving DNA sequences, yet their complexity challenges traditional analytical approaches. This review aims to elucidate how digital technologies such as artificial intelligence (AI) and machine learning (ML) uncover novel epigenetic patterns beyond DNA mutations, enhancing cancer diagnosis, prognosis, and treatment. A comprehensive literature review was conducted, analysing peer-reviewed studies from rom recent reports on PubMed, Scopus, and bioinformatics databases, focusing on AI applications in several cancers such as breast, lung, and colorectal. Findings reveal that ML algorithms, including random forests, convolutional neural networks, and autoencoders, identify epigenetic signatures some of which are methylation patterns and histone modifications with up to 60–90% of cases with high accuracy. These signatures enable early detection, as demonstrated in a pancreatic cancer study achieving 87% sensitivity using ctDNA methylation. AI-driven multi-omics integration has so far uncovered synergistic interactions, improving metastasis prediction by 15–20% in ovarian cancer. Additionally, AI facilitates personalized epigenetic therapies, with models predicting response to histone deacetylase inhibitors in multiple myeloma with 90% accuracy. However, there are limitations to overcome. Some of which are data heterogeneity and model interpretability which exists in the nuances of machine systems. Ongoing research are currently exploring federated learning and explainable AI to address imminent challenges which would enhance generalizability and clinical trust. Future directions include single-cell epigenomics and AI applications in rare cancers to democratize precision medicine. Collaborative efforts among data scientists, clinicians, and biologists are essential to translate these insights into transformative cancer care.
Keywords: Artificial Intelligence (AI), epigenetics, Machine Learning (ML), cancer diagnosis, DNA methylation, histone modifications