Abstract:
The prompt detection of trace amounts of antibiotics is crucial for environmental
surveillance and the entire ecosystem. An electrochemical sensor has been designed to
precisely detect the concentration of ciprofloxacin (CIP) utilizing gold nanoparticles
coated with chitosan on a glassy carbon electrode (AuNPs/CHI/GCE). The influence of
four key electrochemical parameters, viz., scan rate, pH, temperature, and electrode
potential, on the satisfactory detection of CIP was also studied. The accumulated analysis
generated one output parameter, which was current intensity, resulting in 2319 data
points. This dataset was then used in machine learning models to predict the maximum
cathodic current for CIP detection. The best model and its optimal se ings of parameters
were determined after comparing ten different ML models: AdaBoost, Bagging,
CatBoost, Decision Tree, Extra Trees, Gradient Boosting, HistGradientBoosting,
LightGBM, RandomForest, and XGBoost, the performance of the models has been
evaluated by metrics Coefficient of Determination (R²), Mean Square Error (MSE), Root
Mean Square Error (RMSE), Mean Absolute Error (MAE) and Median Absolute Error
(MedAE). It was concluded that CatBoost is the best model both on cumulative
performance and final metrics results. These include an R² score of 0.9985, MSE of 1.468,
RMSE of 1.225, MAE of 0.672, and MedAE of 0.375. Under optimal conditions, the
cathodic current for CIP detection by the CB Model was approximately 128.51 µA, with
a potential of 0.47 V, scanning rate of 98 mV/sec, temperature of 310C, and pH of the
reaction medium at 6.75. These parameters were further used in constructing a
calibration curve using differential pulse voltammetry (DPV), accurately determining
the presence of CIP in known and unknown samples. The LOD, LOQ, and the specific
sensitivity of the developed sensor (AuNPs/CHI/GCE) were 0.049, 0.232 µM, and 786.72
µA µM−1 cm−2, respectively. Finally, a new electrode was fabricated using disposable
materials, specifically from the graphite rod of a dry cell. We evaluated how effectively
the AuNPs-modified Gr electrode sensor could detect traces of CIP in ultra-pure water
samples. The experiment yielded a limit of detection (LOD) of 0.0517 nM and a limit of
quantification (LOQ) of 1.723 nM. The findings indicate that this newly developed
electrode can quickly and accurately identify ciprofloxacin in water samples, making it
valuable for rapid and sensitive analysis.