Escherichia coli O157:H7 survival and transfer dynamics on cold chain packaging materials: An integrated experimental-machine learning framework

Abstract

This study presents a comprehensive investigation of Escherichia coli O157:H7 survival and transfer on six cold chain packaging materials through experimental characterization and machine learning modeling. Survival experiments revealed significant material-dependent variations, with porous PE foamed cotton showing highest bacterial persistence (3.16 ± 0.04 log₁₀(CFU/sample + 1) at 24 h) and smooth nitrile surfaces demonstrating rapid reduction. Transfer rates showed temperature-dependent material ranking, with PE foamed cotton achieving highest efficiency (τ = 0.425 ± 0.021 at 5 °C) and nitrile the lowest (τ = 0.238 ± 0.024). Temperature effects revealed that refrigeration (5 °C) enhanced both bacterial survival and transfer rates (39–107 % increase) compared to freezing (−18 °C), highlighting elevated contamination risks at higher cold chain temperatures. Contact experiments showed instantaneous bacterial transfer with highest efficiency at 2.0 N/cm² within tested pressure range. Machine learning models achieved exceptional performance (R² = 0.9937 for survival, R² = 0.9937 for transfer), with SHAP analysis revealing distinct key determinants: nitrile, time, and temperature for survival prediction, and temperature, time, and surface roughness for transfer prediction. These findings establish the first systematic framework for predicting bacterial behavior on packaging materials and provide evidence-based guidance for cold chain food safety management.