Dynamics of Toxigenesis and Random Forest Prediction Model for Burkholderia gladioli pathovar cocovenenans in Wet...

Food poisoning incidents caused by Burkholderia gladioli pathovar cocovenenans (BGC) in wet rice noodles have raised significant public health concerns. This study investigated the growth of BGC and the production dynamics of BA and TF under five constant temperature conditions, aiming to elucidate the interactive effects of temperature and time on BGC growth and toxin generation, to reveal the dynamic patterns and underlying mechanisms of toxin production, and to preliminarily explore the construction and the application value of a random forest predictive model. The results showed that temperature had a highly significant effect on both BGC growth and BA production (p < 0.001), and a significant effect on TF production (p = 0.015). Time had a highly significant effect on BGC growth and BA production (p < 0.001), but no significant effect on TF production (p = 0.160). The interaction between temperature and time had a highly significant effect on BGC growth and BA production (p < 0.001), but no significant effect on TF production (p = 0.076). At 30 °C, the linear correlation between BA and TF was strongest (p < 0.05). The preliminary predictive models for BA and TF demonstrated good predictive performance (R² 0.955 and 0.789, respectively). Tenfold cross-validation indicated that the internal generalization capability was within an acceptable range, whereas the cross-batch external validation revealed limited generalization capability. These findings provide empirical evidence for understanding the toxigenic mechanisms of BGC and offer a quantitative reference for risk assessment and control of BGC contamination in wet rice noodle matrices.