Forecasting Vibrio Parahaemolyticus Outbreak Risks in Marine Environment

Abstract:

Vibrio parahaemolyticus (V.p) is an epidemiologically significant pathogen that grows naturally in warm marine environment, posing risks to beachgoers and shellfish (primarily oyster) consumers. It is, therefore, important to forecast the occurrence and particularly the infection risk with sufficient lead time so that early warning could be issued in a timely fashion. To that end this paper presents a forecasting model with the lead time of four days for predicting the risk of V.p abundance in marine environment (particularly oysters) by utilizing eight years of field sampling data and the Random Forest method within the R statistical computing environment. While the model input variables include only two independent variables (including Sea Surface Temperature and Sea Surface Salinity), it was found that the V.p abundance is controlled by antecedent environmental conditions. Specifically, the V.p abundance depends on the Sea Surface Temperature of 4, 5 and 8 days before and the Sea Surface Salinity of 4, 7, and 11 days before. Model forecasting results showed that the model was able to correctly predict 85% (model accuracy) of V.p outbreaks. The area under the Receiver Operator Characteristic curve for the model is 0.94, demonstrating the excellent performance of the forecasting model. In addition, an uncertainty analysis was conducted using the bootstrap method to understand the uncertainty involved in model predictions. The uncertainty analysis result (95% confidence band) indicated that the standard error involved in the model predictions is 0.025 that is very small. The Random Forest-based forecasting model enables public health managers to focus more on preventing V.p infections, rather than relying on reacting to problems after they have occurred, greatly reducing the risk of V.p infection to human health and the risk of economic loss to the seafood industry.