A machine learning approach to elucidating physicochemical drivers of opportunistic pathogen occurrence in drinking water systems

Nov 6, 2026   10:00 - 10:50 am  
2311 NCEL - Yeh Center
Sponsor
CEE 595AG - Environmental Engineering & Science
Speaker
Juliana Marks, PhD Student | Advisor Helen Nguyen
Views
41
Originating Calendar
CEE Seminars and Conferences

ABSTRACT
Opportunistic pathogens (OPs) in premise plumbing and drinking water distribution systems (DWDS) can cause severe illness, particularly in immunocompromised individuals. However, effective OP control and risk mitigation require an understanding of the complex interactions among environmental factors that influence OP growth. This study applies machine learning (ML) models to elucidate relationships between physicochemical parameters, OPs (Legionella spp., Pseudomonas spp., and Mycobacterium spp.), and the free-living amoeba Vermamoeba vermiformis, which can harbor OPs and is frequently detected in DWDS.

 Using datasets collected from premise plumbing and DWDS, three ML models, eXtreme Gradient Boosting (XGBoost), Random Forest (RF), and Support Vector Regression (SVR), were optimized by cross-validation to predict OP and V. vermiformis concentrations based on environmental parameters. Data were partitioned into training (80%) and testing (20%) sets to independently evaluate model applicability to unseen data. Model performance was evaluated using the coefficient of determination (R2), root mean squared error (RMSE), and mean absolute error (MAE). Shapley Additive exPlanations (SHAP) was then employed to interpret the influence of each environmental parameter on model output, revealing free chlorine, pH, and temperature as the strongest drivers of Legionella spp. occurrence. SHAP analysis also indicated that Mycobacterium spp. and V. vermiformis were the most influential contributors to one another’s predicted concentrations, likely due to their free chlorine resistance and host-parasite relationship. 

Overall, this study demonstrates the potential of ML frameworks to illuminate complex interactions between OPs and environmental parameters, as well as to inform effective OP control strategies in premise plumbing and DWDS.

SPEAKER BIO
Juliana Marks is a second-year Environmental Engineering and Science MS/PhD student in the Department of Civil and Environmental Engineering at the University of Illinois Urbana-Champaign. Her research, advised by Dr. Helen Nguyen, focuses on understanding opportunistic pathogen and microbial growth in drinking water systems and point-of-use filtration devices, as well as associated public health risks. Juliana received her B.S.E. in Environmental Engineering from the University of Michigan. Prior to graduate school, she worked as a Civil Engineer for the U.S. Army Corps of Engineers, Detroit District.

link for robots only