Researchers develop AI tool to predict E. coli contamination before beaches close
Scientists at the FAMU-FSU College of Engineering have built an artificial intelligence system that predicts unsafe water conditions up to 24 hours before swimmers are exposed to E. coli contamination. The model identifies dangerous levels with approximately 85% accuracy by analyzing rainfall, river flow, turbidity, temperature and upstream conditions in near real time.
Beach closures typically occur after laboratory tests confirm contamination - a process that takes 18 to 24 hours. By then, swimmers may have already been exposed. Assistant Professor Nasrin Alamdari and her team built the framework to shift water quality monitoring from reactive to predictive.
"Our goal is to move from a reactive approach to a predictive one, leveraging continuous environmental data to estimate E. coli levels in near real time and up to a day in advance," Alamdari said.
How the model works
The system ingests current and historical environmental data without waiting for lab results. It combines upstream hydrologic conditions, streamflow rates, rainfall totals, turbidity readings and water temperature to flag elevated contamination risk.
A 2023 sewage spill at the Big Creek Water Reclamation Facility illustrates the problem the model addresses. A sudden treatment failure rapidly contaminated downstream recreational waters. Traditional monitoring would have missed the surge until after people entered the water.
The model accounts for how quickly E. coli levels spike during heavy rainfall - sometimes within hours. It also factors in watershed wetness indicators to improve predictions during moderate rainfall events that standard models often miss.
Health and economic consequences
E. coli infection causes gastrointestinal distress, nausea and fatigue. Young children and older adults face greater risk. Delayed alerts increase medical costs from waterborne illness.
Unexpected closures harm local economies. Hotels, outfitters and water recreation businesses lose revenue with little warning. Municipalities absorb higher costs for emergency notifications and health response. Repeated advisories erode public trust and depress long-term visitation.
Proactive alerts give businesses and government agencies advance notice to reduce unnecessary closures and protect both public health and economic stability.
Urbanization intensifies risk
Between 2007 and 2023, urbanization in the study area increased impervious cover from 24% to 28%, altering runoff pathways and raising E. coli variability in streams. As precipitation patterns grow less predictable, even moderate rainfall carries elevated contamination risk in urban watersheds.
The findings underscore how development decisions influence water quality and public health, pointing to the need for green infrastructure that reduces runoff.
The research was published in Water Research and supported by grants from Florida State University. The work demonstrates how data analysis applied to environmental monitoring can improve community preparedness during contamination events. For those working in water quality, public health or environmental science, this research shows a practical application of predictive modeling to infrastructure challenges.
Your membership also unlocks:
