Limited sampling strategies (LSS) are widely used to estimate surrogates of drug exposure, such as the area under the concentration-time curve (AUC), which generally correlates with drug effect.
The aim of these LSS techniques is to reduce the inconvenient and frequent blood samplings, while keeping the precision of the derived estimates A regression procedure is usually proposed for the prediction of AUC involving only a small number of blood samples collected at specific times as dependent variables. Considering the trade off between the accuracy of the estimation and the clinical convenience, the challenge of LSS is to identify a suitable set of sample concentrations that can achieve this twofold goal. In this presentation, we will discuss the problematic and results around the multiple linear regression with confidence intervals computations, cross-validation methods, and appropriate performance criteria to solve this practical problem. As direct fallout, we have implemented a user friendly tool, with a graphical user interface, that can assist clinicians to set up LSS for their practical needs. In this talk, I will show how this software can be used to efficiently identify the best LSS and the associated statistics, as well as the evaluation of the performance of different LSSs based on various criteria. Such tools are very appreciated in drug development and clinical practice.
This work is in collaboration with J. Li, S. Sarem, F. Nekka and the Pharmacology Unit of Ste-Justine Hospital with C. Litalien, Y. Theoret and A.L Lepayreque.