New highly accurate sensor for E. coli risk detection

Researchers at CU Boulder have developed and validated a new sensor for E. coli risk detection that features an impressive 83% accuracy rate when detecting contamination in surface waters.

Researchers at CU Boulder have developed and validated a new sensor for E. coli risk detection. It features an impressive 83 per cent accuracy rate when detecting contamination in surface waters.

The findings were recently published in Water Research and could improve the detection of a variety of contaminants quickly and effectively in water systems around the globe and America.

Emily Bedell (PhDEnvEngr’22) is the paper’s lead author from the Mortenson Center in Global Engineering. She said about two billion people worldwide use a drinking water source with some level of fecal matter. It can cause health issues, from diarrhoea to stunted growth, especially in young children.

“About 60% of all diarrheal deaths are related to water quality, globally,” she said. “This is a real problem. Current methods for finding poop in drinking water are expensive, high barriers to entry like extensive training or can take about 24 hours to provide results. We have invented a sensor combined with a machine learning model that uses fluorescence to show fecal contamination spikes in real-time.”

Both the sensor and machine learning model combination have been approved for a patent by the U.S. Patent Office.

Bedell said fluorescence works by shining a UV LED light source on a water sample. It measures the amount of light absorbed and re-emitted at a higher wavelength. That information can quickly show potential contamination. However, it is sensitive to many environmental and physical factors such as sample temperature. That can cause noise in the data and make it difficult to interpret.

“We use machine learning techniques to cut through that noise to detect anomalies better,” Bedell said.

Desperate need for e. Coli risk detection

Fast and accurate assessment of water quality is a growing need. It is not restricted to low-income countries but in situations like the Flint, Michigan water crisis.

Professor Evan Thomas, director of the Mortenson Center, is a co-author of the paper. He said climate change is also a factor in this discussion as more frequent power outages may impact treatment facility operations, and severe weather could contaminate critical water sources.

“We are going to need more data on water quality, and we need it to be widely available,” he said. “Taking measurements once a day will not be enough to ensure we are receiving water that doesn’t have biological or chemical contaminants that can harm us in the short and long term.”

Bedell is now employed as an engineer for Virridy in Boulder and is working on advancing the technology further. Ideally, it will partner with a larger home treatment system for those utilising a private well – where water quality is not regulated by the EPA – for their drinking water.

“That sensor will be a miniaturised version of the design built in this paper and will be installed on a house’s main water line coming from the well,” she said. “The sensor’s data will be sent through the user’s WiFi to an online database where the machine learning model will be applied to predict risk level. It will send the information to a mobile app that will alert the user if contamination is detected.”

Future of research

Bedell said she has always been interested in the intersections of engineering, the environment and social equity. This research project brought those aspects together during her time with the Mortenson Center.

“Water quality research hits on all those points in so many ways. With more data, we can explicitly point out how and when communities are being harmed through environmental injustices. It would allows the policies and practices that caused the harm can be addressed,” she said.

Related Articles:

Send this to a friend