Classification of Fish Freshness and Prediction of Mesophilic Aerobic Microbial Count with an Electronic Nose.

The short shelf life of fresh fish and the need to ensure its sanitary acceptability to consumers pose a challenge to the distribution chain. In this scenario, it would be greatly beneficial to develop control tools such as the electronic nose (Enose), capable of classifying fish freshness based on its olfactory pattern and predicting microbial parameters quickly, in real-time, and at a low cost. The E-nose used in this study is a prototype equipped with five low-power metal oxide semiconductor (MOS) gas sensors, portable, and connected via Bluetooth to a mobile application. E-nose measurements were processed using discriminant neural network analysis (ANNDA), achieving an 85% success rate in classifying fish freshness. Additionally, a predictive correlation model was obtained, using partial least squares analysis (PLS), between the mesophilic aerobic bacterial count of the fish and the assessment of its olfactory pattern with the E-nose, yielding R
²
values ≥ 0.95 and RMSECV values ≤ 0.43. The results of this research suggest that the E-nose is suitable for predicting the number of mesophilic aerobes and classifying fish according to their freshness status.