An iterative curve stripping technique for pharmacokinetic parameter estimation

Non-linear least-squares regression is commonly used for pharmacokinetic parameter estimation. Initial parameter estimates are required as a prelude to non-linear least-squares and the quality of the final parameter estimates may depend on these initial values. Polyexponential curve stripping is frequently used for the provision of initial estimates. It is demonstrated that under certain conditions conventional curve stripping yields biased estimates. A new iterative curve stripping technique is developed and is shown to be free of such bias. The two methods are compared using both simulated and real pharmacokinetic data.