Intelligent colorimetric gas-sensing film based on a pH-responsive hydrogel for beef freshness evaluation via machine learning

Continuous meat freshness monitoring is vital for food safety and waste reduction. Conventionally, colorimetric pH sensors detect spoilage based on volatile basic nitrogen compounds but often require liquid-phase analysis and invasive sampling. In this study, we developed an intelligent colorimetric gas-sensing system based on a pH-responsive hydrogel, featuring phenol red covalently grafted onto a chitosan backbone and crosslinked with poly(vinyl alcohol). The hydrogel film exhibits a distinct yellow-to-purple transition upon exposure to ammonia gas. To quantify freshness, machine learning models (XGBoost and MLP) were implemented and refined via Bayesian optimization. These models accurately decoded the nonlinear relationships between colorimetric responses and freshness indicators, predicting total viable count ( R 2 = 0.94) and total volatile basic nitrogen ( R 2 = 0.99) with high precision. This digitized platform enables non-invasive assessment, offering significant potential for next-generation smart packaging.