Machine learning in understanding environmental variability of vibriosis in coastal waters

<i>Vibrio spp.</i> comprise ecologically significant bacteria that thrive in warm, moderately saline water, and their incidence and proliferation are strongly influenced by environmental factors. In recent years, <i>Vibrio spp</i>. infections have been reported more frequently and over a greater geographical area along the US eastern seaboard. This study provides an analysis of <i>Vibrio spp</i>. infections, notably caused by <i>Vibrio alginolyticus</i>, <i>Vibrio cholerae</i> non-O1/non-O139, <i>Vibrio fluvialis</i>, <i>Vibrio mimicus</i>, <i>V. parahaemolyticus</i>, and <i>V. vulnificus</i>, extracted from the Centers for Disease Control and Prevention's Cholera and Other <i>Vibrio</i> Illness Surveillance system, located within 200 km of the eastern US coast, to analyze latitudinal distribution trends between 1990 and 2019. For each <i>Vibrio spp</i>., case incidence (presence/absence) was modeled, including environmental data and employing extreme gradient boosting machine learning algorithms (XGBoost). Environmental parameters associated with the incidence of vibriosis were clustered using k-means clustering. The northern limit of total cases of vibriosis was found to have increased <i>ca</i>. 40 km/year, with <i>V. alginolyticus</i> (<i>ca</i>. 70 km/year), <i>V. fluvialis</i> (<i>ca</i>. 60 km/year), and <i>Vibrio parahaemolyticus</i> (<i>ca</i>. 60 km/year) showing the greatest latitudinal shifts. These changes were found to be linked to environmental parameters that enhance the proliferation of <i>Vibrio spp</i>. The average accuracy of the XGBoost models was 60.9%-71.0%, with temperature and salinity being the most significant predictors. Relationships among other environmental parameters were complex and nonlinear, but phytoplankton and precipitation served to differentiate the models. Clustering using k-means yielded results that supported temperature, salinity, and phytoplankton as important environmental parameters. Research in progress will aid in developing global predictive risk models for <i>Vibrio spp</i>. infections.IMPORTANCE<i>Vibrio spp</i>. are ecologically significant bacteria, and their incidence and proliferation are strongly influenced by environmental factors. In recent years, <i>Vibrio spp</i>. infections have been reported more frequently and over a greater geographical area along the US eastern seaboard. This study provides an analysis of latitudinal distribution trends of <i>Vibrio spp</i>. infections, notably caused by <i>Vibrio alginolyticus</i>, <i>Vibrio cholerae</i> non-O1/non-O139, <i>Vibrio fluvialis</i>, <i>Vibrio mimicus</i>, <i>Vibrio parahaemolyticus</i>, and <i>V. vulnificus</i>, within 200 km of the eastern US coast. The northern limit of total cases of vibriosis was found to have increased <i>ca</i>. 40 km/year. These changes were found to be linked to environmental parameters that enhance the proliferation of <i>Vibrio spp</i>. Temperature and salinity were the most significant predictors of vibriosis case presence and absence. Phytoplankton and precipitation changes served to differentiate <i>Vibrio sp</i>. presence. Research in progress will aid in developing global predictive risk models for <i>Vibrio spp</i>. infections.