[Study on the Application of NAS-Based Algorithm in the NIR Model Optimization].

Area of Science:

• Analytical Chemistry
• Chemometrics
• Spectroscopy

Background:

• Ginkgo biloba extracts contain multiple components, complicating quantitative analysis.
• Traditional spectral analysis methods face challenges with complex mixtures and unknown interferences.
• Developing robust methods for analyzing active compounds like flavonol aglycones is crucial.

Purpose of the Study:

• To introduce and evaluate Net Analysis Signal (NAS) for preprocessing multi-component Ginkgo biloba leaf extracts.
• To optimize Near-Infrared (NIR) model building using NAS-based two-dimensional correlation analysis.
• To establish simultaneous quantitative models for quercetin, kaempferol, and isorhamnetin.

Main Methods:

• Utilized NAS algorithm for spectral preprocessing.
• Applied NAS-based two-dimensional correlation analysis for wavenumber selection.
• Developed quantitative models using Partial Least Squares (PLS) with NAS-reconstructed spectra.
• Employed Lorber and Goicoechea and Olivieri (HLA/GO) algorithms for NAS vector calculation.

Main Results:

• Successfully established simultaneous quantitative models for quercetin, kaempferol, and isorhamnetin.
• Demonstrated that NAS preprocessing reduced the number of factors and improved model robustness.
• Identified useful spectral regions for model building via NAS-based 2D correlation spectroscopy.
• Validated NAS as an effective tool for spectral preprocessing and model calibration.

Conclusions:

• NAS-based preprocessing methods are effective for analyzing complex mixtures like Ginkgo biloba extracts.
• NAS significantly enhances the accuracy and robustness of quantitative models in NIR spectroscopy.
• The study highlights the practical applicability of NAS for analyzing complex phytochemicals and petrochemical medicines.