|| Checking for direct PDF access through Ovid
Simple, quick, inexpensive and non-destructive technique, it does not employ organic solvents, thus protecting the environment.The models developed by NIR in association with multivariate analysis provided good prediction of the APIs for the external samples.The tablets led to results that were not statistically similar, despite having prediction errors considered acceptable in the literature.PAT technology: models using the same amount of excipients contained in the tablets need to be developed.The World Health Organization recommends that TB treatment be administered using combination therapy. The methodologies for quantifying simultaneously associated drugs are highly complex, being costly, extremely time consuming and producing chemical residues harmful to the environment. The need to seek alternative techniques that minimize these drawbacks is widely discussed in the pharmaceutical industry. Therefore, the objective of this study was to develop and validate a multivariate calibration model in association with the near infrared spectroscopy technique (NIR) for the simultaneous determination of rifampicin, isoniazid, pyrazinamide and ethambutol. These models allow the quality control of these medicines to be optimized using simple, fast, low-cost techniques that produce no chemical waste. In the NIR – PLS method, spectra readings were acquired in the 10,000–4000 cm−1 range using an infrared spectrophotometer (IRPrestige – 21 – Shimadzu) with a resolution of 4 cm−1, 20 sweeps, under controlled temperature and humidity. For construction of the model, the central composite experimental design was employed on the program Statistica 13 (StatSoft Inc.). All spectra were treated by computational tools for multivariate analysis using partial least squares regression (PLS) on the software program Pirouette 3.11 (Infometrix, Inc.). Variable selections were performed by the QSAR modeling program. The models developed by NIR in association with multivariate analysis provided good prediction of the APIs for the external samples and were therefore validated. For the tablets, however, the slightly different quantitative compositions of excipients compared to the mixtures prepared for building the models led to results that were not statistically similar, despite having prediction errors considered acceptable in the literature.