The goals of this research were to test the hypotheses that a new method, NIR/ARS, identifies meclizine and methyltestosterone tablets more accurately than near-IR spectrometry alone, and that bootstrap principal components provide important information on the reliability of wavelengths in principal-component regression with spectra. The near-IR spectra of the two types of tablets are different and they can usually be distinguished by inspection. However, in a manufacturing environment, automated inspection requires equipment that does not fatigue, and or which probabilities of Type I and Type II errors can be estimated. The bootstrap algorithms employed in this study produce z-scores for estimation of Type I and Type II errors, and locate minute distinctions between two samples quickly and with great accuracy. Using these algorithms to confirm the identity of chemical compounds with near-IR and NIR/ARS techniques should provide advantages in terms of time, expense and accuracy.
Near-IR and the new NIR/ARS are among the few methods that can be used nondestructively and with precision for analysis of almost any kind of dosage form or formulation. Acoustic-resonance spectra identify samples using physical properties like density, acoustic-velocity and impedance, while near-IR spectrometry makes use of overtones and combinations of molecular motions. When principal-axis transformation is employed, ARS tends to model major sources of the total spectral variation, freeing near-IR spectra for modeling minor variations (an advantage for analysis of biological materials, where water can be a source of many major components). The combined NIR/ARS technique often gives better results than near-IR spectrometry alone. The chance of accidental contamination or mislabeling of these two drugs (or in general any drugs) can be minimized by an on-line 100% inspection technique. Given proper equipment, every tablet or dosage form could be subjected to quantitative and qualitative analysis using near-IR spectrometry and NIR/ARS as it is manufactured.