Electronic Nose Application for the Determination of Penicillin G in Saanen Goat Milk with Fisher Discriminate and Multilayer Perceptron Neural Network Analyses

Antibiotics are routinely added to milk products and pose potential harm to public health. The objective of this study was to use an innovative and nondestructive application of an electronic nose instrument for rapid detection of penicillin G in goat milk. The PEN3 electronic nose system was utilized to detect volatile substances in goat milk after the addition of penicillin G sodium salt at concentrations of 0, 50, 100 and 200 μg/L. The data were extracted at 60 s to carry out a linear discriminant analysis. Additional statistical analysis was conducted using neural networks to predict the penicillin G concentration in goat milk samples. Accuracy rates for the two methods were 98.0 and 96.7% for training samples, and 97.0 and 94.9% for testing samples, respectively. The results from this study show that the electronic nose system can be utilized to predict the penicillin G concentrations in goat milk samples. PRACTICAL APPLICATIONS: Antibiotics are routinely added to animal‐derived food products and have been reported to cause potential harm to human health. However, traditional analytical chemical methods (i.e., gas, liquid and high‐performance liquid chromatography) can detect trace amounts of additives and have some drawbacks, such as complicated operation, high costs of implementation and lengthy analysis time. This research has examined the feasibility, accuracy and effectiveness of a metal oxide semiconductor gas sensor type electronic nose device (the PEN3 e‐nose) to detect and discriminate among different concentrations of penicillin G in goat milk. Data from this study show that the electronic nose system can be used to predict the penicillin G concentration in goat milk samples.