Coupling of NIR Spectroscopy and Chemometrics for the Quantification of Dexamethasone in Pharmaceutical Formulations

Abstract

Counterfeit or substandard drugs are pharmaceutical formulations in which the active pharmaceutical ingredients (APIs) have been replaced or ingredients do not comply with the drug leaflet. With the outbreak of the COVID-19 pandemic, fraud associated with the preparation of substandard or counterfeit drugs is expected to grow, undermining health systems already weakened by the state of emergency. Analytical chemistry plays a key role in tackling this problem, and in implementing strategies that permit the recognition of uncompliant drugs. In light of this, the present work represents a feasibility study for the development of a NIR-based tool for the quantification of dexamethasone in mixtures of excipients (starch and lactose). Two different regression strategies were tested. The first, based on the coupling of NIR spectra and Partial Least Squares (PLS) provided good results (root mean square error in prediction (RMSEP) of 720 mg/kg), but the most accurate was the second, a strategy exploiting sequential preprocessing through orthogonalization (SPORT), which led (on the external set of mixtures) to an R2pred of 0.9044, and an RMSEP of 450 mg/kg. Eventually, Variable Importance in Projection (VIP) was applied to interpret the obtained results and determine which spectral regions contribute most to the SPORT model.

Quality study of animal‐derived traditional Chinese medicinal materials based on spectral technology: Calculus bovis as a case

“…content of bilirubin, using the following three variable selection methods: competitive adaptively reweighted sampling (CARS), 26 . 27 After using the CARS method, the prediction accuracy of the model was significantly improved, as can be seen from Table 3. The R 2 of the content model was 0.979, which is close to 1, and the root mean square error of calibration (RMSEC) was 2.3515.…”

“…This may be due to the obvious differences in bilirubin among different varieties of C. bovis, while the background complexity caused by other components interfered with the effective information.Therefore, we use three variable selection methods, namely competitive adaptive reweighted sampling (CARS), projected variable importance (VIP), and correlation coefficients (CC). Therefore, we carried out the extraction of important feature wave variables for the content of bilirubin, using the following three variable selection methods: competitive adaptively reweighted sampling (CARS), 26 27 . After using the CARS method, the prediction accuracy of the model was significantly improved, as can be seen from Table 3.…”

Optical Insights into Fibrotic Livers: Applications of Near-Infrared Spectroscopy and Machine Learning

“…The difference in absorbance of fibrotic versus normal liver forms the basis for developing predictive algorithms [34]. Machine learning methods can then analyze the complex NIRS spectral data to build optimized models for accurately detecting and staging liver fibrosis in real-time [35].…”

A Feasibility Study towards the On-Line Quality Assessment of Pesto Sauce Production by NIR and Chemometrics

“…It has been extensively demonstrated how Multivariate Statistical Process Monitoring/Control based on Latent Variables (MSPC-LVs) can lead to an efficient process monitoring [ 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 ].…”