Predictive water virology using regularized regression analyses for projecting virus inactivation efficiency in ozone...

Area of Science:

• Environmental Science
• Public Health
• Water Treatment Engineering

Background:

• Wastewater reclamation and reuse are vital for water-stressed regions.
• Insufficiently treated wastewater poses risks due to waterborne pathogens.
• Sanitation Safety Planning (SSP) uses Hazard Analysis and Critical Control Points (HACCP) to manage these risks.

Purpose of the Study:

• To develop predictive models for virus log reduction values (LRVs) in wastewater disinfection.
• To identify key water quality and operational parameters influencing disinfection efficacy.
• To support the determination of appropriate critical limits (CLs) for safe wastewater reuse.

Main Methods:

• Utilized five machine learning algorithms to model virus LRVs.
• Employed water quality and operational parameters as explanatory variables for ozone disinfection.
• Evaluated model performance using datasets outside the training range.

Main Results:

• Automatic relevance determination with interaction terms showed superior prediction for norovirus and rotavirus LRVs.
• Bayesian ridge with interaction terms and lasso with quadratic terms accurately predicted poliovirus and coxsackievirus LRVs, respectively.
• Models demonstrated robustness in predicting LRVs for new datasets.

Conclusions:

• The developed models provide a framework for projecting virus LRVs in wastewater treatment.
• These models can assist wastewater treatment plant operators and risk assessors in setting safe critical limits.
• Further data collection is recommended to enhance model predictability and flexibility for diverse wastewater scenarios.