A Biopsychosocial Risk Score for Stratifying Disease Vulnerability in Healthy Populations: A Prospective Cohort and Multi-Omics Study in the UK Biobank

Abstract Proactive identification of systemic vulnerability for disease(s) before clinical onset in healthy individuals is an ultimate goal of preventive and precision medicine, yet current tools remain largely disease-specific and fail to quantify latent vulnerability, an integrative measure of underlying health status, for early prevention and risk-stratified intervention. To address this, we developed the Risk Score for Disease Vulnerability (RS4DV) based on 85 accessible biopsychosocial measures, which was constructed using a Light Gradient Boosting Machine trained and validated in the UK Biobank ( n = 391,193). Its capacity to capture pre-clinical vulnerability was subsequently evaluated in a held-out cohort free of baseline diagnoses ( n = 35,193). Over a median follow-up of 14.7 years, baseline RS4DV stratified long-term health outcomes in the held-out cohort, in which high-risk individuals exhibited accelerated disease accumulation (HR = 2.53, 95% CI: 2.44-2.62) and elevated risk of all-cause mortality (HR = 4.03, 95% CI: 3.68-4.41). Multi-omics analyses further revealed that RS4DV captures signatures of systemic inflammation, metabolic dysregulation, and accelerated brain ageing, establishing its biological interpretability. To facilitate real-world translation, we developed a practically feasible version of the RS4DV with only six routinely accessible items, which maintained robust predictive fidelity relative to the full model. This light-version model demonstrated robust generalizability in the 1970 British Birth Cohort under a zero-shot paradigm. Collectively, RS4DV provides a biologically grounded and scalable tool for personalized risk assessment decades before clinical onset and early-stage risk stratification, enabling a paradigm shift toward proactive health management and precision prevention.