AI-Powered Nutrigenetics Slashes BMI and Reverses Biological Aging in Morbid Obesity

Morbid obesity — defined as BMI ≥40 kg/m² — carries genetic heritability estimates exceeding 80%, yet most clinical interventions remain one-size-fits-all. This Croatian study addressed that gap by integrating whole-genome sequencing (WGS), biochemical panels, hormonal assays, and plasma glycomics to generate individualized dietary plans for 14 consecutively recruited adults with morbid obesity, then tracked outcomes over six months using a patented unsupervised machine learning clustering platform (German Patent Office No. DE 20 2025 101 197 U1).

At enrollment, the cohort displayed a classic metabolic storm: elevated fasting glucose, hyperinsulinemia, high HOMA-IR scores indicative of insulin resistance, raised triglycerides, and elevated C-reactive protein signaling chronic low-grade inflammation. The GlycanAge index — a validated biomarker of biological age derived from IgG N-glycan profiles — averaged 56 ± 12.45 years, while mean chronological age was only 32 ± 9.62 years, a 24-year gap reflecting profoundly accelerated biological aging driven by the pro-inflammatory glycomic signature of obesity.

Genomic analysis revealed that the majority of participants carried risk alleles in the fat mass and obesity-associated gene (FTO) and the interleukin-6 gene (IL-6), both well-established drivers of adiposity and systemic inflammation respectively. Beyond these headline genes, clinically significant variants were identified across more than 40 genes implicated in metabolic regulation, nutrient metabolism, cardiometabolic risk, hereditary cancer predisposition, and nutritional status. The AI-driven unsupervised clustering algorithm processed these polygenic profiles and stratified all 14 patients into a single coherent cluster, suggesting a shared molecular phenotype despite surface-level variation in individual risk variants — a finding that underscores the convergent biology of severe obesity.

Personalized dietary recommendations generated by the platform were followed for six months. The primary anthropometric outcome was dramatic: mean BMI fell from 52.09 ± 7.41 to 34.6 ± 9.06 kg/m² (p <0.01), representing a reduction of approximately 33.6% and moving the average participant from Class III to Class I obesity. Crucially, GlycanAge also declined significantly, from 56 ± 12.45 to 48 ± 14.83 years (p <0.01), an 8-year reduction in biological age within just six months. This parallel improvement in both anthropometric and glycomic aging markers suggests that the intervention addressed underlying inflammatory and metabolic dysregulation, not merely caloric balance.

The clinical implications are notable for both patients and practitioners. GlycanAge as a response biomarker offers a measurable, biologically meaningful endpoint beyond BMI, potentially motivating sustained adherence. The finding that nutrigenetic profiling across 40+ genes can generate actionable dietary guidance — and that following it produces large effect sizes in a short timeframe — supports broader adoption of precision nutrition frameworks. Caveats are significant: the sample size of 14 patients with no randomized control group limits causal inference, and the lack of an external validation cohort means the machine learning clusters require independent replication before clinical deployment.