Sustainable analysis of COVID-19 Co-packaged paxlovid: exploring advanced sampling techniques and multivariate...

Area of Science:

• Analytical Chemistry
• Chemometrics
• Computational Statistics

Background:

• Random sampling methods present limitations in accuracy and predictive validity for complex datasets.
• Developing robust and efficient analytical methods is crucial for pharmaceutical analysis and quality control.
• Existing chemometric models require optimization for enhanced data coverage and predictive performance.

Purpose of the Study:

• To integrate Latin Hypercube Sampling (LHS) with various multivariate chemometric models (PLS, GA-PLS, ANN, MCR-ALS).
• To enhance the predictive accuracy and robustness of these models using LHS for improved data sampling.
• To demonstrate the application and benefits of LHS in analyzing the anti-COVID-19 drug Paxlovid (ritonavir-boosted nirmatrelvir).

Main Methods:

• Integration of Latin Hypercube Sampling (LHS) with Partial Least Squares (PLS), Genetic Algorithm-Partial Least Squares (GA-PLS), Artificial Neural Networks (ANN), and Multivariate Curve Resolution-Alternating Least Squares (MCR-ALS).
• Application of LHS to sample selection for analyzing ritonavir (RNV) and nirmatrelvir (NMV) in Paxlovid.
• Development of a hybrid variable selection strategy (GA-ICOMP-PLS) and assessment of eco-friendly analytical metrics (Sample Preparation Metric of Sustainability, Analytical Greenness metrics, RGB12, Blueness Applicability Grade Index).

Main Results:

• LHS provided well-interpreted samples, capturing essential variability without increasing sample size, outperforming random sampling.
• ANN and MCR-ALS models showed significant Root Mean Square Error of Prediction (RMSEP) reductions (14.1-53.1%) for RNV and NMV.
• The GA-ICOMP-PLS model achieved low prediction errors (0.15 for RNV, 0.14 for NMV), demonstrating strong predictive power and generalization.
• Eco-friendly assessments indicated high sustainability scores and broad applicability, aligning with green analytical chemistry principles.

Conclusions:

• Latin Hypercube Sampling (LHS) is a powerful technique for enhancing the robustness and predictive accuracy of chemometric models.
• The developed models and sampling strategies offer efficient and reliable analysis of pharmaceutical compounds like Paxlovid.
• The study highlights the successful implementation of sustainable and eco-friendly analytical practices in pharmaceutical analysis.