|| Checking for direct PDF access through Ovid
An efficient method for designing an NIR calibration set in spectral space is demonstrated for quantitative analysis of pharmaceutical tablets.The new calibration set required fewer samples compared to a traditional full factorial calibration set in order to predict a model drug concentration in different sets of tablet.Similar (if not improved) model robustness can be achieved using fewer samples by designing calibration set in spectral space.Designing a calibration set is the first step in developing a multivariate spectroscopic calibration method for quantitative analysis of pharmaceutical tablets. This step is critical because successful model development depends on the suitability of the calibration data. For spectroscopic-based methods, traditional concentration based techniques for designing calibration sets are prone to have redundant information while simultaneously lacking necessary information for a successful calibration model. A method for designing a calibration set in spectral space was developed. The pure component spectra of a tablet formulation were used to define the spectral space of that formulation. This method maximizes the information content of measurements and minimizes sample requirements to provide an efficient means for developing multivariate spectroscopic calibration. A comparative study was conducted between a commonly employed full factorial approach to calibration development and the newly developed technique. The comparison was based on a system to quantify a model drug, acetaminophen, in pharmaceutical compacts using near infrared spectroscopy. A 2-factor full factorial design (acetaminophen with 5 levels and MCC:Lactose with 3 levels) was used for calibration development. Three replicates at each design point resulted in a total of 45 tablets for the calibration set. Using the newly developed spectral based method, 11 tablets were prepared for the calibration set. Partial least square (PLS) models were developed from respective calibration sets. Model performance was comprehensively assessed based on the ability to predict acetaminophen concentrations in multiple prediction sets. One prediction set contained similar information to calibration set while the other prediction sets contained different information from calibration set in order to assess the model accuracy and robustness. Similar prediction performance was achieved using the 11-tablet design (spectral space), compared to the 45-tablet full factorial approach. This work demonstrates that a calibration set designed in spectral space provided an efficient means of developing spectroscopic multivariate calibration.