# Optimization of SERS Detection for Sulfathiazole Residues in Chicken Blood Using GA-SVR

## TL;DR

A new SERS method with GA-SVR optimization is developed to detect sulfathiazole residues in chicken blood efficiently and accurately.

## Contribution

A novel GA-SVR model integrated with preprocessing techniques is introduced for enhanced sulfathiazole residue detection in complex matrices.

## Key findings

- Optimized SERS parameters achieved high SERS intensities at 1157 cm−1 for sulfathiazole detection.
- GA-SVR model demonstrated strong predictive performance with R2p of 0.9278 and RMSEP of 3.1552.
- The method showed high specificity and minimal matrix interference in chicken blood analysis.

## Abstract

The extensive use of sulfathiazole in poultry farming has raised growing concerns regarding its residues in poultry-derived products, posing risks to human health and food safety. To overcome the limitations of conventional detection methods and address the analytical challenges posed by inherent complexity of chicken blood matrix for the detection of sulfathiazole residues in chicken blood, a rapid and sensitive surface-enhanced Raman spectroscopy (SERS) method was developed for detecting sulfathiazole residues in chicken blood. Four colloidal substrates, i.e., gold colloid A, gold colloid B, gold colloid C, and silver colloids, were synthesized and evaluated for their SERS enhancement capabilities. Key parameters, including electrolyte type (NaCl solution), colloidal substrate type (gold colloid A), volume of gold colloid A (550 μL), volume of NaCl solution (60 μL), and adsorption time (14 min), were systematically optimized to maximize SERS intensities at 1157 cm−1. Furthermore, a genetic algorithm-support vector regression (GA-SVR) model integrated with adaptive iteratively reweighted penalized least squares (air-PLS) and multiplicative scatter correction (MSC) preprocessing demonstrated superior predictive performance with a prediction set coefficient of determination (R2p) value of 0.9278 and a root mean square error of prediction (RMSEP) of 3.1552. The proposed method demonstrated high specificity, minimal matrix interference, and robustness, making it suitable for reliable detection of sulfathiazole residues in chicken blood and compliant with global food safety requirements.