Nonlinear Parameter and State Estimation Approach in End-stage Kidney Disease Patients

BackgroundBlood and fluid volume management in End-stage Kidney Disease (ESKD) patients plays an essential role in dialysis therapy to replace kidney function. Reliable knowledge of blood and fluid volumes before and during dialysis could be used to improve treatment outcomes significantly. ObjectiveThis study aims to develop an estimation approach providing predictable information on blood and fluid volumes before and during a regular dialysis routine. MethodsA new approach is developed to estimate blood volume, fluid overload, and vascular refilling parameters from dialysis data. The method utilizes a nonlinear fluid volume model, an optimization technique, and the Unscented Kalman Filter (UKF) incorporated with data. This method does not rely on restricted ultrafiltration (UF) and dilution protocols and uses the Fisher information matrix to quantify error estimation. ResultsAccurate estimations for blood volumes (5.9{+/-}0.07L and 4.8{+/-}0.03L) and interstitial fluid volumes (18.81{+/-}0.15L and 12.19{+/-}0.03) were calculated from dialysis data consisting of constant and stepwise UF profiles. We demonstrated that by implementing the estimated parameters into the model, a precise prediction of the measured hematocrit (HCT) can be achieved during the treatment. ConclusionWe showed that the result does not depend highly on initial conditions and can be accurately estimated from a short data segment. A new method, applicable to the current dialysis routine, is now available for ESKD patients to be implemented within the dialysis machines.