Linear parameter varying system based modeling of hemodynamic response to profiled hemodialysis

This paper proposes a novel linear parameter varying (LPV) system to model the hemodynamic response of end-stage renal failure patients to profiled hemodialysis (PHD). Ultrafiltration rate (UFR) and dialysate sodium concentration (Na) are imposed as the control inputs and the model computes the relative blood volume (RBV), percentage change in heart rate (ΔDHR(%)) and percentage change in systolic blood pressure (ΔDSBP(%)) during the course of hemodialysis. Model parameters are estimated using least squares approach based on data collected from 12 patients where each patient underwent 4 profile hemodialysis sessions. Parameter identification based on four profiled sessions of the same patient revealed an average mean square error of 0.11 for RBV, 0.24 for ΔDHR and 0.43 for ΔDSBP. The results provided a good model to estimate the individual patient's hemodynamic behavior during hemodialysis. The developed model can play a vital role in designing a robust control system to automatically regulate the UFR and Na while maintaining the hemodynamic variables within stable range.