Using robust Bayesian network to estimate the residuals of fluoroquinolone antibiotic in soil

摘要: Prediction of antibiotic pollution and its consequences is difficult, due to the uncertainties complexities associated with multiple related factors. This article employed domain knowledge spatial data construct a Bayesian network (BN) model assess fluoroquinolone (FQs) in soil an intensive vegetable cultivation area. The results show: (1) relationships between FQs contributory factors: Three factors (cultivation methods, crop rotations, chicken manure types) were consistently identified as predictors topological structures three FQs, indicating their importance pollution; deduced knowledge, methods are determined by which require different nutrients (derived from manure) according plant biomass. (2) performance BN model: integrative robust achieved highest detection probability (pd) high-risk receiver operating characteristic (ROC) area, since it incorporates uncertainty. Our encouraging findings have implications for use approach assessment informing decisions on appropriate remedial measures.