Individually identifiable model for process optimization in clinical dialysis

Optimal of f line process control of dialysis with respect to improved adequacy requires a comprehensive computer model of the exchange mechanisms during dialysis. An extended multicompartment model for the exchange of sodium, potassium, chloride, urea, creatinine, hydrogen ions, bicarbonate, CO_2 has been developed. Additionally osmotic water shifts between different distribution spaces and the resting membrane potential of excitable cells are calculated. The influence of hemodynamic effects has been considered by modeling 2 differently perfused parallel tissue groups. The model consists of two subsystems with 8 and 5 individually adjustable parameters. Sensitivity analysis has been carried out with a finite difference algorithm. Conditional local identifiability has been proven for a simplified version with 6 and 3 parameters, which have been estimated from measured data with a simplexalgorithm. First validation of the model has been performed successfully with measured data from 18 chronic hemodialysis patients. Alterations of cardiac output and tissue perfusion on the dynamics of sodium, potassium and the acid base state have been found to exert minor influence on the results as long as no rapid processes (bolus infusions, interruptions) are simulated. Extensions towards modeling of the calcium kinetics and hemodiafiltration have been done.