Estimation d'une droite d'étalonnage à partir d'un seul point de gamme par la méthode bayésienne: Application au dosage de l'azidothymidine

✍ Scribed by M Tod; P Nicolas; O Petitjean

• Publisher: Elsevier Science
• Year: 1992
• Tongue: English
• Weight: 686 KB
• Volume: 260
• Category: Article
• ISSN: 0003-2670
• DOI: 10.1016/0003-2670(92)80130-y

✦ Synopsis

Linear calibration using a single calibration stanakrd and bayesian method. Application to azidothymidine determination.) An empirical bayesian method for processing linear calibration data using a single calibration standard is described. An exact solution to the bayesian estimator has been obtained and allowed the calculation of the bayesian estimate of the calibration parameters with no need of any software. The procedure was validated through the determination of bias and precision of axidothymidine liquid chromatographic assay. The error in predicting analyte concentrations using bayesian estimation of the calibration line ranged from -0.3 to -3.1%, and the root-meansquare error ranged from 7.8 to 14.2%. Performances of the familiar least-squares method using six calibration standards were determined, and were very similar to those of the bayesian method. The influence of several experimental errors and violation of the hypothesis upon which the bayesian method lies were also discussed. Thus, the method described in this paper is time-saving and should be most useful when quick results are needed for a small number of samples, as in clinical toxicology.

Resum6

L'utilisation de la methode bayesienne pour estimer une droite d'Ctalonnage B partir dun seul point de gamme est d&rite. Une solution exacte au critere d'optimisation bayesien a CtC obtenue et permet de mettre en oeuvre la mtthode bayesienne sans recourir a un outil informatique. La solution propode a et6 valid&e en determinant l'exactitude et la pr&ision du dosage de l'axidothymidine par chromatographie liquide. Le biais de l'estimation baCsienne Ctait compris entre -0,3 et -3,1%, tandis que la precision &tit comprise entre 7,8 et 14,2% selon les concentrations ?I mesurer. Ces performances ont 6t6 comparees B celles de l'etalonnage classique en six points de gamme et ont trouvees t&s voisines. L'influence de differentes erreurs experimentales et du non respect des hypothbses sur lesquelles repose la mCthode baygsienne a egalement et6 d&cut&e. Ainsi, la methode bayCsienne permet une 6conomie importante sur le temps d'analyse lorsqu'il faut doser un petit nombre d'Cchantillons, en rdduisant la durie de 1'Ctalonnage.