Biologically Informed Machine Learning Prioritizes Dietary Supplements That Protect Neural Crest Cells from Ethanol-Induced Epigenetic Dysregulation and Developmental Impairment.

The impairment of neural crest cells (NCCs) plays a pivotal role in the pathogenesis of fetal alcohol spectrum disorders (FASD). Epigenetic regulators mediate ethanol-induced disruptions in NCC development and represent promising targets for nutritional interventions. Here, we developed a biologically informed machine learning framework to predict nutritional supplements that modulate five key epigenetic regulators (miR-34a, DNMT3a, HDAC, miR-125b, and miR-135a) and mitigate ethanol's adverse effects on NCCs. The optimized models demonstrated robust predictive performance and identified a number of nutritional supplements that could attenuate ethanol-induced NCC impairment, including resveratrol, vitamin B12, emodin, quercetin, and broccoli sprout-derived compounds. Our optimized models also revealed structural features that are critical for mitigating ethanol-induced NCC impairment through specific epigenetic mechanisms. These findings support predictive modeling as a tool to prioritize nutritional supplements for further investigation and the development of dietary strategies to prevent or reduce the risk of FASD.