Global regression model for moisture content determination using near-infrared spectroscopy

Area of Science:

• Pharmaceutical Sciences
• Analytical Chemistry
• Spectroscopy

Background:

• Accurate moisture content (MC) determination is critical for freeze-dried drug product stability and quality.
• Traditional methods like Karl Fischer (KF) titration can be time-consuming for in-process or release testing.
• Near-infrared (NIR) spectroscopy offers a potential rapid, non-destructive alternative for MC analysis.

Purpose of the Study:

• To develop and evaluate a global quantitative near-infrared (NIR) model for determining moisture content (MC) in diverse freeze-dried pharmaceutical products.
• To assess the performance of various chemometric algorithms for NIR-based MC quantification.
• To validate the optimized global model for accuracy, linearity, and risk-based assessment.

Main Methods:

• Acquisition of 3822 NIR spectra from three different freeze-dried drug products using two spectrometers to capture variability.
• Development and comparison of linear and non-linear regression models, including Partial Least Square (PLS), Decision Tree (DT), Bayesian Ridge Regression (Bayes-RR), K-Nearest Neighbors (KNN), and Support Vector Regression (SVR).
• Reference MC determination using the Karl Fischer (KF) method, with values ranging from 0.05% to 4.96%.

Main Results:

• The Support Vector Regression (SVR) algorithm yielded the best performing global model for MC determination.
• The optimized SVR model achieved a Standard Error of Calibration (SEC) of 0.12% and a Standard Error of Prediction (SEP) of 0.15%.
• The global NIR model demonstrated acceptable accuracy, linearity, and risk-based performance for simultaneous MC determination.

Conclusions:

• A single, global NIR quantitative model can effectively determine moisture content across different freeze-dried pharmaceutical products.
• This approach significantly accelerates validation time and in-lab release analyses compared to traditional methods.
• NIR spectroscopy presents an innovative and efficient tool for quality control of freeze-dried drug products.