Obesity affects over 650 million people globally and drives metabolic disorders such as type 2 diabetes (T2DM) and cardiovascular diseases (CVD), which cause 17.9 million deaths annually. Multi-omics approaches spanning genomics, transcriptomics, proteomics, and metabolomics combined with artificial intelligence (AI) offer powerful tools to unravel obesity-related disease mechanisms and improve prediction models. Recent studies show that applying machine learning to high dimensional omics data can identify biomarkers and improve prediction accuracy by 5 to 15% over the 85 to 90% baseline achieved in similar early detection tasks. Deep learning models capture complex patterns in heterogeneous data but face challenges in harmonizing diverse omics types. This review critically evaluates AI-driven multi-omics integration strategies, compares their strengths, limitations, and optimal use cases, and outlines key considerations such as data standardization, privacy, and regulatory compliance for clinical translation in obesity research.
Saikia et al. (Tue,) studied this question.