A Continuous, In-situ, Near-time Fluorescence Sensor Coupled with a Machine Learning Model for Highly Accurate Detection of Fecal Contamination in Drinking Water: Design, Characterization, and Field Validation

Bedell, Emily

Graduate Thesis Or Dissertation

A Continuous, In-situ, Near-time Fluorescence Sensor Coupled with a Machine Learning Model for Highly Accurate Detection of Fecal Contamination in Drinking Water: Design, Characterization, and Field Validation Public Deposited

Analytics

Download PDF

Citations

Citeable URL: https://scholar.colorado.edu/concern/graduate_thesis_or_dissertations/b8515p864

Abstract

Equitable access to reliable, affordable, and safe drinking water is essential to human health and livelihood. Globally, two billion people use a drinking water source that is contaminated with feces. Low-cost, field-deployable, near-time methods for assessing water quality are not available when and where waterborne infection risks are greatest. In this dissertation, I describe the development and testing of a novel device for the measurement of online, in-situ, and remotely reporting tryptophan-like fluorescence (TLF), making use of recent advances in deep-ultraviolet light emitting diodes (UV-LEDs) and sensitive silicon photomultipliers. TLF is an emerging indicator of microbial water quality that is associated with members of the coliform group of bacteria and therefore potential fecal contamination. After optimizing the sensor's sensitivity to 0.05 ppb tryptophan, I demonstrated the close correlation between TLF and E. coli in model waters and proof of principle with sensitivity of 33 0mL for lab grown E. coli and 10 CFU/100mL for E. coli in wastewater.

I characterized the sensor's behavior to multiple fluorescence quenching parameters through benchtop analysis. Fluorescence response declined with water temperature and a correction factor was calculated. Inner filter effects were shown to have negligible impact in an operational context. Biofouling was demonstrated to increase the fluorescence signal by approximately 82%, while mineral scaling reduced the sensitivity of the sensor by approximately 5%. A training and validation data set for a machine learning model was built by installing four sensors on Boulder Creek, Colorado for 88 days and enumerating 298 grab samples for E. coli with membrane filtration. The machine learning model incorporated a proxy feature for fouling (time since last cleaning) which improved model performance. The model was able to predict high risk fecal contamination with 83% accuracy (95% CI: 78% - 87%), sensitivity of 80%, and specificity of 86%. A model distinguishing between all World Health Organization established risk categories performed with an overall accuracy of 64%. The sensor design combined with the highly skilled model has the ability to provide water service providers as well as individual consumers more reliable and informative data about fecal contamination risk in their drinking water. Findings to date suggest that this device represents a scalable solution for remote monitoring of drinking water supplies to identify high-risk fecal contamination in drinking water in near-time. Such information can be immediately actionable to reduce risks and would reduce cost of microbial testing greatly, improving health and wellness of consumers and enabling water service providers more access to funds that can be used to increase access to clean water.

Creator