Multivariate calibration of spectrophotometric data using a partial least squares with data fusion

Area of Science:

• Analytical Chemistry
• Spectrophotometry
• Chemometrics

Background:

• Simultaneous spectrophotometric determination of multiple analytes, such as Cu(II), Ni(II), and Cr(III), often faces challenges due to spectral overlap.
• Traditional methods like partial least squares (PLS) regression can be limited in extracting characteristic information and achieving high regression quality in complex mixtures.
• Data fusion (DF) techniques offer a way to integrate information from diverse sources, enhancing model performance.

Purpose of the Study:

• To develop and validate a novel hybrid method, data fusion combined with partial least squares (DF-PLS), for enhanced simultaneous spectrophotometric determination.
• To improve the extraction of characteristic information and the quality of regression for multicomponent analysis.
• To demonstrate the superiority of the DF-PLS method over conventional PLS for analyzing samples with severe spectral overlap.

Main Methods:

• A novel DF-PLS method was developed, integrating data fusion principles with partial least squares (PLS) regression.
• Wavelet transforms were employed for noise reduction and feature extraction across different scales, treating each scale as a distinct information source.
• Optimization of wavelet functions (Daubechies 4), decomposition level (7), thresholding method (HYBRID), and PLS factors (3) was performed.

Main Results:

• The DF-PLS method achieved significantly lower relative standard errors of prediction (RSEP) of 3.13% compared to 10.3% for the standard PLS method.
• Successful simultaneous determination of Cu(II), Ni(II), and Cr(III) was demonstrated, even in the presence of severe spectral overlap.
• The DF-PLS method proved to be more effective than the conventional PLS method for this analytical task.

Conclusions:

• The DF-PLS method effectively combines the strengths of data fusion and PLS regression, offering enhanced capabilities for spectrophotometric analysis.
• This hybrid approach provides a promising and attractive solution for simultaneous multicomponent determination, particularly in challenging spectral matrices.
• The DF-PLS technique represents a significant advancement in chemometric applications for spectrophotometric analysis.