ABSTRACT Golden pomfret, with its abundant production, ease of capture, tender and flavorful flesh, and rich content of unsaturated fatty acids, has gained widespread popularity among consumers. However, with the development of the society, the traditional meat freshness testing methods can't satisfy the requirements of modern fisheries for rapid, nondestructive and accurate testing of aquatic products. Gas‐sensor technology can achieve rapid and nondestructive determination of freshness by detecting volatile odors in aquatic products. However, the sensor is easily influenced by the interference of environmental factors due to its high sensitivity. How to eliminate the interference factors and realize the rapid detection of aquatic product quality has become a new research direction. Therefore, we analyze the basic working mechanism of the gas sensor based on the target molecule's adsorption characteristics. After the preprocessing such as filtering and de‐noising the original response curve of the sensor, we propose a feature value extraction method which is more suitable for the adsorption principle of the gas sensor. The method combines the mechanical features of the sensor adsorption process so that the feature values extracted can more accurately reflect the nature of the interaction between the target gas and the sensor surface, thereby enhancing the sensitivity of the model to freshness. We applied this feature extraction method to obtain three feature value parameters, and then we used various machine learning algorithms to build the golden pomfret freshness prediction model. The comparative analysis of these models showed that the model based on the Random Forest algorithm achieved an accuracy of 0.776 ± 0.041, a precision of 0.751 ± 0.040, a recall of 0.721 ± 0.065, and an F 1‐score of 0.722 ± 0.056 under five‐fold cross‐validation. The model implementation has been basically satisfied to predict the golden pomfret freshness. Our study provides a methodological reference for feature extraction in gas‐sensing applications and supports other research on quality evaluation of golden pomfret and other aquatic products.
LI et al. (2026) studied this question.