AI-Powered Precision Medicine and Sustainable Prevention for Chronic Disease Biomarkers and Patient-Reported Well-Being Outcomes

Abstract

AI-directed precision medicine promises to revolutionize the management of chronic disease from a reactive approach. This study created and/or tested an AI-based precision-prevention model using a synthetic longitudinal dataset of 320 people randomly randomised to either AI-assisted precision prevention versus routine preventive therapy over the course of 16 weeks. Data were evaluated using mixed-effects models, time-variant CM biomarkers, anthropometric measures, lifestyle behaviours, the WHO-5 Well-being scale or EQ-5D-5L utility measure, and EQ-VAS. For model making, Shapley additive explanations (SHAP), regularised logistic regression (LR), random forest (RF), and extreme GBM (XGBoost) predicted clinical response. AI-assisted prevention improved HbA1c by −0.221% compared to conventional treatment, whereas body weight and BMI decreased (−2.440kg/m² vs-0.655), with greater reductions in FBS (−7.174 mg/dL), systolic blood pressure (−4.589mmHg), LDL-C (−8.374 mg/dL), and triglycerides (−11.498mg/dL, p<0.05). The WHO-5 increased (6.420), as did the EQ-5D-5L (0.041) and EQ-VAS score (5.120). AI-guided therapy, 65.6% of subjects achieved clinical outcomes in the intervention arm versus 37.5% in the standard care group. XGBoost had the highest classification performance (ROC-AUC = .783). These simulated results demonstrate the analytic power of integrating explainable AI, precision preventive biomarkers, and patient-reported outcomes but need to be validated on clinical value using real-world clinical data.