Raman Spectroscopy and Feature Selection Partial Least Squares Regression (PLSR) for the Determination of the anti-Diabetic Drugs Vildagliptin and Metformin in Combination Therapy

The present study focused on quantifying the active pharmaceutical ingredients (APIs) metformin and vildagliptin in their fixed-dose combination product using Raman spectroscopy with manual feature selection and partial least squares regression (PLSR). Raman spectroscopy has become an important technique for qualitative and quantitative analysis of pharmaceuticals. This study assesses its potential for analyzing solid dosage combinations of the two anti-diabetic drugs, vildagliptin and metformin. We prepared different solid dosage forms by mixing the APIs with excipients. After gathering and preparing the Raman spectra, these were evaluated by mean plot analysis, principal component analysis (PCA), and manual feature selection-based PLSR. Mean plot analysis of pure APIs and excipients also helped us select the features manually for PLSR. Through the use of multivariate analysis such as PCA or this PLSR approach, the study demonstrates how Raman spectroscopy can accurately measure the concentration of APIs (such as vildagliptin and metformin) with high goodness-of-fit (R2) values (near 0.90). Its low root mean square error of calibration (RMSEC) and root mean square error of validation (RMSEV) for predictions of blind samples showcase its high accuracy for quantitative analysis. These findings make Raman spectroscopy a highly feasible technique for drug formulation analysis, providing a rapid, low-cost alternative to traditional techniques and suggesting that it be more widely adopted in industry quality control and drug development.