Finite sampling design with quantization Applied to test of the bioavaibility

Kamal Boukhetala (1) , Karim Benhenni (2) , Zohra Habib (1)

1 Faculté des Mathématiques
2 IPS - Inférence Processus Stochastiques

The optimal sampling problem for the estimation of an integral, based on observations, with random and correlated measurement errors is treated. Considering the effect of the quantization of errors, we determine the optimal design, which is constructed by utilizing stochastic algorithms. The efficiency of the Genetic Algorithm is confirmed by comparison with other implemented algorithms, by considering a pharmacokinetic application for a test of bioavailability. A multiobjective approach is proposed for the choice of an optimal compromise design between the variability of the quantization of errors and the integral's estimator..

Mots clés: Bioavailability; Optimal sampling; Quantization; Genetic Algorithm; Annealing Genetic Algorithm; Multiobjective Evolutionary Algorithms

Domaines: Méthodologie [stat.ME]

Kamal Boukhetala, Karim Benhenni, Zohra Habib. Finite sampling design with quantization Applied to test of the bioavaibility. 2011. ⟨hal-00619141⟩