Drug Designing: Open Access
Open Access

Selection of best calibration model during bioanalytical method validation

International Conference and Expo on Drug Discovery & Designing

August 11-13, 2015 Frankfurt, Germany

Sandeep Sonawane

Posters-Accepted Abstracts: Drug Des

Abstract :

The quality of bioanalytical data is highly dependent on using an appropriate regression model. The most common and simplest
approach to fit a calibration curve to data points (x, y) is by ordinary linear regression and to express the correlation coefficient
(r2) to define the degree of association between these two variables with the acceptable value greater than 0.99. However, r2 alone
is not adequate to demonstrate linearity, as the ordinary linear regression approach presupposes that each data point in the selected
range has a constant absolute variable (i.e. homoscedasticity). But most of the bioanalytical assays usually have to cover a broad
concentration range and the variance is more likely to increase with concentration (i.e. heteroscedasticity) which finally impairs
accuracy despite of acceptable r2 value. An alternative approach is to use the weighted linear regression which normally generates a
better curve fit than ordinary linear regression and increases accuracy over the whole concentration range. So there is a need to study
and suggest an approach in selection of a calibration model during bioanalytical method validation which offers degree of assurance
that the developed method will hold true during routine use.