Quantitative analysis of dried serum FTIR spectra based on correlation Analysis-Interval random Frog-Partial least...

Area of Science:

• Biochemistry
• Analytical Chemistry
• Spectroscopy

Background:

• Serum biochemical markers are crucial for clinical diagnostics but often involve costly reagents and lengthy analysis times.
• Infrared spectroscopy presents a reagent-free, rapid, and simultaneous multi-parameter analysis alternative for serum.
• Current methods for serum component quantification can be expensive and time-consuming.

Purpose of the Study:

• To explore the relationship between dried serum infrared spectra and biochemical parameters.
• To investigate the feasibility of simultaneously quantifying nine major serum components using dried serum infrared spectra.
• To develop a novel, rapid, and accurate method for real-time serum component determination.

Main Methods:

• Collected serum samples from 66 healthy subjects.
• Employed correlation analysis to identify relevant spectral bands for glucose, protein, and lipids.
• Utilized the Interval Random Frog (IRF) algorithm to select optimal wavenumbers and constructed Partial Least Squares (PLS) quantitative models (CA-IRF-PLS).

Main Results:

• Successfully quantified nine major serum components, including glucose, total protein, lipids, and lipoproteins, with high accuracy.
• Achieved correlation coefficients (Rp) in the test set ranging from 0.8892 to 0.9941.
• The developed CA-IRF-PLS method demonstrated superior performance compared to conventional PLS and SPA-PLS methods.

Conclusions:

• The CA-IRF-PLS method enables rapid and accurate quantification of multiple serum components from dried serum infrared spectra.
• This technique offers a promising, cost-effective, and efficient alternative for clinical diagnostics.
• The study presents a novel approach for real-time determination of clinical parameters in serum.