AI-Assisted Nutrigenomic Modeling of Blueberry Anthocyanin–Gut Microbiome Interactions to Predict Butyrate Production and Insulin Sensitivity Improvement in Type II Diabetes

Abstract

Background: Type II diabetes mellitus (T2DM) is characterized by insulin resistance, chronic inflammation, impaired glucose metabolism, and gut microbial dysbiosis. Emerging evidence suggests that anthocyanin-rich berries can modulate gut microbial composition and functionality, leading to enhanced production of beneficial metabolites such as butyrate. Integrating nutrigenomics, microbiome science, metabolomics, and artificial intelligence offers a promising approach for identifying personalized nutritional interventions and predictive biomarkers for metabolic health. Objective: This study aimed to investigate the effects of blueberry anthocyanins on gut microbiome composition, butyrate production pathways, and insulin sensitivity, while developing artificial intelligence-based predictive models for individualized response assessment in Type II diabetes. Methods: Freeze-dried blueberry powder standardized for anthocyanin content was incorporated into dietary interventions at four levels: T₀ (0%), T₁ (5%), T₂ (10%), and T₃ (15%). A simulated 12-week randomized intervention involving 120 adults with Type II diabetes was designed. Gut microbiome profiling was performed using shotgun metagenomic sequencing, while metabolomic analyses quantified short-chain fatty acids and microbial metabolites. Nutrigenomic assessment included transcriptomic analysis of IRS1, AKT1, GLUT4, PPARγ, AMPK, CPT1A, and inflammatory markers. Multi-omics integration was conducted using MOFA+, DIABLO, and WGCNA. Machine learning models including Random Forest, XGBoost, and LightGBM were developed to predict butyrate production and insulin sensitivity outcomes. Results: Blueberry anthocyanin supplementation significantly increased Akkermansia muciniphila, Faecalibacterium prausnitzii, Roseburia spp., and butyrate synthesis genes. T₂ demonstrated optimal efficacy with a 118% increase in fecal butyrate concentration, 32.6% reduction in HOMA-IR, 0.81% reduction in HbA1c, and significant upregulation of IRS1, GLUT4, and AMPK signaling pathways. Multi-omics analyses identified gut microbial abundance, anthocyanin metabolites, and butyrate production genes as major determinants of insulin sensitivity improvement. XGBoost achieved the highest predictive performance for identifying responders (ROC-AUC = 0.95). Conclusion: Blueberry anthocyanins significantly improved insulin sensitivity through modulation of gut microbial ecology and enhancement of butyrate production. AI-assisted nutrigenomic modeling successfully identified predictive biomarkers and personalized response patterns, highlighting the potential of precision nutrition strategies for Type II diabetes management.