Modeling and forecasting vibrio vulnificus concentration of long-range dependence on marine environmental conditions

Vibrio vulnificus (vvh) is an epidemiologically significant bacterium that naturally occurs in coastal waters under favorable environmental conditions and causes one of the highest mortality rates among known foodborne pathogens. Little is currently known about the environmental conditions that favor the abundance of vvh particularly in seafood like oysters. By synergistically combining 13 years of field oyster sampling-based vvh concentration data from environmentally distinct coastal waters, satellite remote sensing-based environmental data, and the Artificial Intelligence-based machine learning technique, this paper presents new insights into the favorable environmental conditions and new ensemble models for forecasting the vvh concentration in oysters. Specifically, it was discovered that the concentration of vvh in oysters is controlled by antecedent conditions of seven environmental predictors, including solar radiation (SR), water level (WL), wind speed (WS), chlorophyll a (CHL), pH, water temperature (SST), and salinity (SSS). The key novelty of this work is the discovery of SR as being the most important predictor of vvh despite its high correlation with SST. WL is the second (or third) most important predictor of vvh. It was also discovered that there is a long-range dependence of vvh concentration on antecedent conditions of the seven environmental predictors up to 80 days before. This is another novel contribution of this work. Based on the findings, four XGBoost ensemble models with varying lead times were presented for forecasting the vvh concentration in oysters, providing a scientific basis and sufficient response time for proactive management interventions to protect public health from potential vvh infections.