Numéro de la revue: 26
Auteurs: Hayet Merabet
Laboratoire de Mathématiques Appliquées et Modélisation, Département de Mathématiques, Université Mentouri – Constantine Route de Ain-El Bey, 25000 Constantine, Algérie
This work is a generalization and systematization of the methodology for clinical trials in a Bayesian framework. We have used a purely Bayesian sequential aspect. This article provides a solution fully Bayesian that incorporates the prevision in a global issue. In an intermediate analysis, the predictive inference focuses on all the data, the data available and future data, in this manner, the evaluation of the prevision error is not overvalued as in an approach that does take into account the future observation. We applied the proposed procedures to the Gaussian model and it was possible to reach an explicit form of the various probabilities of errors that the practitioner can make. Thus we can make available to the user an implementable tool and fully Bayesian.
The sequential aspect of the treatment adopted in this paper is a particularly innovative element compared to existing technology; it also helps to reduce multiphase studies more ambitious than the existing, which for the patient makes the analysis more ethical since it allows a stoppage of the experience shorter and less tardy.
Key Words: Predictive methods-Bayesian analysis-Clinical Trials-p-Value.