Towards multiscale modelling of placental oxygen transport using capillary micro-cells
Please login to view abstract download link
The placenta supplies oxygen to a growing baby, connecting two separate blood systems (one from the mother and one from the fetus) that do not mix. Oxygen exchange happens across fetal capillaries, and the shape and distribution of these vessels strongly affect the efficiency of oxygen exchange and delivery. Existing computer models either capture these small details with high computational cost or simplify them, losing important features on capillary architecture. This study presents a physiology-based approach to model oxygen flow in realistic but simplified synthetic micro-cell networks that mimic fetal capillary structures. Using statistical data from real placental capillaries, we created two-dimensional models that reproduce variability in capillary length, diameter and branching patterns. We then simulated blood flow and oxygen transport under different inlet flow rates and oxygen levels, creating over 2,500 examples to study how network structure affects oxygen delivery. We defined oxygen network efficiency as how closely the oxygen gained in the simulated blood flow matched the maximum possible uptake. Machine learning methods (XGBoost with SHAP analysis) were used to assess which features most strongly influenced efficiency. The inlet flow rate had the biggest effect, while vessel size, resistance, and branching also played important roles. Faster flow reduced efficiency because blood spent less time in contact with surrounding oxygen, whereas larger diameters and more branched networks improved distribution and exchange. Overall, this work provides a practical modelling framework that captures detailed micro-scale behaviour while remaining efficient enough to apply in larger-scale simulations incorporating more complex fetoplacental vascular structures.
