Multi-Omics and AI-/ML-Driven Integration of Nutrition and Metabolism in Cancer: A Systematic Review, Meta-Analysis, and Translational Algorithm
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Cancer is increasingly recognized as a metabolic disease with strong nutritional determinants. Recent advances in multi-omics technologies and artificial intelligence (AI), especially machine learning (ML), have enabled novel integrative frameworks to decode complex interactions among diet, metabolism, and tumor biology.
Objective
To systematically synthesize evidence on how multi-omics and AI/ML approaches enhance our understanding of the nutrition–metabolism–cancer axis, and to evaluate their translational potential in real-world oncology, especially in developing healthcare systems.
Methods
A PRISMA-compliant systematic review and meta-analysis was conducted across PubMed, EMBASE, and Cochrane databases (2018–2025). Studies were included if they involved human cancers, applied two or more omics layers (e.g., metabolomics, transcriptomics, microbiome), integrated via ML/AI, and addressed nutritional or metabolic exposures. Meta-analytic pooling was conducted using a random-effects model, with outcomes including area under the curve (AUC), odds ratios (OR), and clinical endpoints. Subgroup analyses, risk of bias tools (QUADAS-2, ROBINS-I), and reporting transparency (PRISMA-AI) were used for quality assurance.
Results
From 4,812 records, 42 studies met the inclusion criteria. The pooled AUC was 0.81 (95% CI: 0.78–0.84) and OR was 2.4 (95% CI: 1.2–3.5), indicating robust diagnostic and predictive capability. Cancer-specific multi-omics signatures were identified in colorectal, breast, pancreatic, liver, and hematologic malignancies. Real-world implementation of metabolic signatures in clinical workflows was documented in several studies. Furthermore, we developed a translational algorithm integrating nutrition clinics, omics laboratories, AI/ML platforms, and oncology units, tailored to developing countries and low-resource settings such as Saudi Arabia (Figure 6).
Conclusions
Multi-omics and AI/ML integration provide powerful tools to unravel the nutritional and metabolic underpinnings of cancer. Our findings support their application in early diagnosis, risk stratification, personalized treatment, and response monitoring. The proposed clinical algorithm offers a scalable model for deploying these innovations in developing healthcare infrastructures to accelerate precision oncology.
