Maxym Shmatkov1, Janko Diminic1, Jurica Zucko1, Olga Gornik2, Valentina Borko2, Toma Keser2, Andrija Karačić3, Robert Keser4 and Antonio Starcevic1*
1University of Zagreb Faculty of Food Technology and Biotechnology
2University of Zagreb Faculty of Pharmacy and Biochemistry
3Centar Mikrobiom d.o.o.
4In silico d.o.o.
astar [at] pbf.hr
Abstract
Metabolic syndrome represents a progressive continuum of systemic dysfunction, yet current diagnostic paradigms rely on late-stage clinical biomarkers (e.g., fasting glucose, HOMA-IR) that fail to capture latent metaflammation. Furthermore, standard population-level multi-omics analyses are frequently confounded by immutable physical and demographic baseline variances. Here, we present a graph-based, synthetic digital twin architecture applied to a well-characterized multi-omic clinical cohort (N=300). By employing a K-nearest neighbor (K-NN) sex-stratified matching algorithm that strictly anchors patients to their exact physical baselines, we mathematically isolated pure metabolic pathophysiology from background biological noise.
Applying this computational framework, we identified a highly prevalent latent-risk group—traditionally misclassified as healthy by standard clinical criteria—that exhibits near-complete phenotypic overlap with the manifest disease cohort. We demonstrate that this latent metabolic decline is characterized and driven by profound shifts in the glycosylation profile of Complement Component 3 (C3), a central protein in the innate immune system. Principal component analysis isolated a discrete C3-glycan inflammatory axis dominated by five specific glycoforms (CIIIGIRMN1N2H10, CIIIGIRMN1N2H8, CIIIGIRMN1N2H9, CIIIKTVLT1N2H5, and CIIIKTVLT1N2H6). Crucially, this C3-glycosylation shift precedes traditional dysglycemia and correlates directly with downstream systemic immune activation and targeted gut microbiome perturbations, specifically the depletion of butyrate-producing Lachnospirales.
To translate these high-dimensional omics findings into clinical utility, we integrated an AI-assisted semantic ontology to generate a deterministic Clinical Decision Support System (CDSS). This engine autonomously separates structural psychological traits from actionable physiological behaviors, generating personalized, biologically realistic intervention targets derived directly from a patient’s healthy synthetic twins. Ultimately, our findings establish specific C3-glycan modifications as critical, early-stage biomarkers of metabolic decline, and validate the synthetic digital twin network as a robust, scalable framework for precision preventive medicine.
Keywords: Digital twin, glycomics, metaflammation
Acknowledgement: This research was funded by the European Union (NextGenerationEU) project “The glycome and microbiome as markers of dietary impact on the health of women of reproductive age” within the program ‘Targeted Scientific Research’ (NPOO.C3.2.R3-I1.04.0073). The work presented has been filed as a EU patent application EP26173207.7 “METHOD AND SYSTEM FOR EARLY DETECTION OF LATENT METABOLIC RISK BASED ON COMPLEMENT C3 GLYCOSYLATION AND HEALTHY-TWIN RECOMMENDATION ANALYSIS.”