Machine learning framework enables low-cost water toxin detection
Machine learning framework enables low-cost water toxin detection
Publish Date: 2026-07-11 07:36:00
Source Domain: www.news-medical.net
-
Portable screen-printed carbon electrode (SPCE) biosensors are a rapid and low-cost method for detecting the toxic microcystin-lysine-arginine (MC-LR) produced by cyanobacteria during harmful algal blooms in freshwater.
-
SPCE sensors work by measuring electrochemical changes corresponding to MC-LR concentration; however, they can be inaccurate due to interference from varying water quality parameters like pH and turbidity.
-
Researchers from Hanbat National University and the University of Central Florida developed a machine learning framework to improve the accuracy of SPCE sensors for MC-LR detection, without requiring repeated recalibration for various water samples.
-
The machine learning model, specifically Extreme Gradient Boosting (XGBoost), excelled in predicting MC-LR concentrations across diverse water conditions with high accuracy.
-
The model, trained with 201 measurements from 27 field sites, identifies critical water quality parameters influencing the prediction, with the biosensor’s electrical impedance being the strongest predictor.
-
This framework is significant as it reduces the need for sample-specific calibration, cutting down on time, labor, and sensor consumption, thereby lowering costs and improving analytical efficiency.
-
This advancement in biosensor technology and machine learning offers a practical and effective approach for on-site detection of MC-LR in various environmental waters, especially important as harmful algal blooms increase due to climate change.
-
The collaborative work led by Professors Jungsu Park and Woo Hyoung Lee, published in Volume 298 of the journal Water Research, represents a step forward in environmental monitoring and public health protection against water-borne toxins.
