Explainable multiomic screening for suboptimal health status in young adults: 3PM-guided innovation is envisaged

Suboptimal health status (SHS) is a reversible predisease stage and represents a key "window of opportunity" for predictive, preventive, and personalized medicine (3PM/PPPM). However, current screening methods still rely mainly on subjective questionnaires and lack objective, interpretable, and actionable tools for timely intervention. We aimed to develop an exploratory prototype system that combines multiomic signals with explainable artificial intelligence to apply 3PM in young adults.

Transcriptomic, metabolomic, and gut microbiome data from 30 SHS patients and 35 healthy controls were analyzed. Seven machine learning algorithms were compared, with elastic net selected for its balance of accuracy, stability, and interpretability. Calibration and decision curve analyses were performed to test robustness and clinical utility. Shapley additive explanations (SHAP) were applied for both global and individual interpretations. The multiomic elastic net prototype achieved high and stable discrimination (accuracy 0.941, ROC-AUC 0.999), with strong calibration and net benefit. Beyond statistical performance, the system identified biologically plausible and modifiable molecular targets-such as reduced vitamin K and elevated glycerophosphocholine-that are directly amenable to preventive strategies. SHAP further provided individual-level profiles, revealing the specific biological drivers of SHS risk for each participant and offering a template for personalized recommendations.

This study proposes an innovative 3PM-guided prototype system for predicting suboptimal health status on the basis of multiomics data. We suggest embedding this tool into preventive healthcare to enable early risk prediction, applying personalized interventions to delay or reverse the progression of SHS, and providing individualized follow-up to support long-term health management. From a public health perspective, this approach may substantially reduce the future burden of chronic diseases by addressing risks at a reversible stage.

The online version contains supplementary material available at 10.1007/s13167-025-00426-3.