Bayesian Estimation for Model Parameters and Time Delay of Blood Pressure Response to Phenylephrine Drug Infusion

Patient blood pressure response to phenylephrine (PHP) drug infusion is modeled as intra-and-inter-variability model with time delay. This paper proposes a novel Bayesian-filtering-based approach for real-time estimating the model parameters variability and drug transport delay (time delay). From data-driven perspective, cross-correlation (XCF) of mean arterial pressure (MAP) measurement with drug infusion provides prior knowledge of time delay estimation, and cubature Kalman filter (CKF) supplies nonlinearity estimation of model parameters and posterior knowledge of time delay. The Bayesian-filtering-based approach was validated in sequential mechanism with animal experimental data. Real-time experimental test convinced the superiority of the proposed approach than multiple-model EKF (MMEKF) in both estimation accuracy of MAP and drug transport delay estimation and sensitivity analysis of model parameters and drug transportation delay estimation.