Physics-Informed Digital Twin of Maternal-Fetal Hemodynamics for Predictive Risk Simulation in Preeclampsia

Preeclampsia (PE) is a hypertensive disorder of pregnancy characterized by endothelial dysfunction, impaired uteroplacental perfusion, and high maternalfetal morbidity. Although modern biomarkers provide early diagnostic potential, few predictive systems connect molecular imbalance to physical flow abnormalities. Using our previous results, we developed a Physics-Informed Digital Twin that simulates uteroplacental hemodynamics under real physiological constraints. The twin integrates mean arterial pressure, soluble fms-like tyrosine kinase-1, placental growth factor, hepatic function, and neonatal outcomes. The NavierStokes equations govern uterine blood-flow dynamics, while a Physics-Informed Neural Network updates model parameters from clinical inputs. Simulated uterine-artery resistance indices correlated strongly with Doppler-measured values (R2 = 0.88, p < 0.001). The model reproduced the association between high sFlt-1/PlGF ratios (> 85), MAP > 110 mmHg, and increased rates of preterm birth (> 70 %) and neonatal intensive-care unit admissions (> 60 %). The PI-DT links angiogenic imbalance to measurable hemodynamic dysfunction, enabling interpretable maternalfetal risk prediction.