Researchers are developing AI-enabled tools to predict flesh-eating waterborne pathogens

In the summer of 2015, just a few hours after Florida native Justin Grubich noticed a light abrasion on his elbow, he knew something was terribly wrong. 

“I woke up in the middle of the night, and my elbow had blown up, like a pimple coming off of it,” he remembered. “I thought, oh, I got just a little bit of an infection or something, and I tried to go back to sleep. But it started throbbing and throbbing. By about 6 a.m., my arm had just blown up.” 

Hours later, Grubich was in the hospital, with medical personnel digging deep to evacuate the dying flesh. The suspicion: infection by Vibrio vulnificus, commonly referred to as flesh-eating bacteria. 

And while Grubich lived to tell the tale, University of Florida researchers are working to prevent such close calls. Supported by a $3.6 million grant from the National Institutes of Health, the collaborative project between UF and the University of Maryland will develop an environmentally sensitive pathogen warning system for Vibrio species.  

Led by UF, the initiative will create forecasting tools that can anticipate health risks linked to environmental conditions, providing advance warnings similar to the way weather forecasting helps communities prepare for storms and other hazards.  

At UF’s Department of Environmental Engineering Sciences, or EES, Antarpreet Jutla, Ph.D., is working to ensure that all coastal denizens and inhabitants have the information they need on the dangerous waterborne bacteria. 

As part of a groundbreaking research initiative supported by the NIH, Jutla and his collaborator, Katherine Deliz Quiñones, Ph.D., also from EES, are working to transform the way scientists detect, forecast and respond to dangerous waterborne infections caused by pathogenic Vibrio bacteria in coastal waters.  

The research team ultimately hopes to create a real-time public health information platform capable of delivering timely, science-based risk alerts and water quality information to government agencies and the public. 

The project will develop VIGOR, or Vibrio Infection Genomics for Outbreak Risk Forecasting, an AI-enabled satellite-based forecasting platform designed to identify environmental conditions associated with increased risk of Vibrio exposures, particularly during and after algal blooms and extreme weather events such as flooding following hurricanes. 

Grubich had most likely been exposed to a pathogenic species of Vibrio bacteria, which includes Vibrio vulnificus, Vibrio parahaemolyticus and Vibrio alginolyticus. These naturally occurring marine bacteria are found in coastal waters and can cause severe and sometimes life-threatening infections through the consumption of contaminated seafood or exposure of open wounds to seawater.  

Their growth and transmission are strongly tied to environmental conditions such as sea surface temperature, salinity, nutrient availability and ecological disturbances. 

“Vibrio (species) are ubiquitous in coastal waters. There are hundreds of species,” Deliz Quiñones explained. “Dozens of them can be pathogenic to humans and marine creatures, including fish. That's why we need to monitor — detection alone doesn't answer our questions.” 

The forecasting component of VIGOR will integrate satellite remote sensing data, environmental parameters, microbial genomic sequencing and machine learning-based ecological niche algorithms to predict seasonal and geographic hotspots for pathogenic Vibrio bacteria.  

These predictive models will also incorporate demographic and infrastructure vulnerability indicators to identify communities facing the greatest public health risks. 

“So, the short-term goal for the next four to five years is to have a validated early warning model,” Jutla explained. “But the long-term goal is to be able to understand how the microbiome of coastal waters behaves over seasons, species and spaces.” 

The team is asking: What is the relative abundance of pathogens in the water? Do they change over time? How do species variations occur? Do they have preferential moving patterns? They want to position UF as the national leader for microbial water research and information.

As Jutla began delving into hydrological modeling of water-borne diseases, his team was running into a problem: a dearth of data.   

Enter Deliz Quiñones.

“We're collaborating to generate the microbiological and water quality data that they will be integrating in the models,” she said. 

Data collection will happen in the old-fashioned way — powered by student researchers in the field.

The water-quality measurements generated by Deliz Quiñones’ lab will be used to train the models to identify microbial indicators that signal deteriorating water quality or heightened disease risk before outbreaks occur. 

Building on experience gleaned from a similar project that focused on Vibrio cholerae, Jutla hopes to build a VIGOR dashboard interface, web-based, that can project risky water conditions for all Vibrio species for three to four weeks ahead of time. 

 “Our goal is to increase public awareness of emerging environmental risks and provide people with the information they need to make informed decisions,” Jutla said.