Dynamic Modelling of Obstetric Patient Coagulation from Kaolin-Activated Thromboelastogram Data

Abstract Defects in blood clotting (coagulopathies) are linked to severe outcomes in mothers suffering from obstetrical hemorrhage. Identifying patients with a coagulopathy poses a challenge for clinicians, who are required to make quick treatment decisions in fast-paced environments with a high degree of uncertainty. Integrating data from point-of-care coagulation tests with mathematical models of coagulation presents an exciting opportunity to improve patient outcomes by reducing this uncertainty. A model with parameters estimated from individual patient data can provide clinicians with a way to compare patients and group them into categories of probable coagulopathy based on biologically-interpretable parameters. With this in mind, we developed a mechanism-inspired model of blood coagulation calibrated against thromboelastogram (TEG) data. Markov Chain Monte Carlo and sensitivity analysis were used to assess the identifiability and distribution of model parameters for 25 obstetric patients. The ability of our model to separate patients in parameter space based on differences in observed TEG response lends credence to the feasibility of using dynamic models as tools for identifying coagulopathy subtypes within the obstetric population.