Wastewater Surveillance with Machine Learning: Insights from Queensland SARS-CoV-۲ Data

Wastewater-based epidemiology (WBE) has emerged as a powerful tool for monitoring community health and providing early warnings of infectious disease outbreaks. During the COVID-۱۹ pandemic, WBE was widely implemented to complement clinical testing and inform public health interventions. This study examines the Queensland SARS-CoV-۲ wastewater surveillance dataset and explores its potential not only for retrospective monitoring but also for predictive modeling. After data cleaning and feature engineering, we applied Random Forest and XGBoost classifiers to predict the likelihood of detecting SARS-CoV-۲ in wastewater samples. Despite the limited number of labeled observations, both models achieved high performance, with accuracies exceeding ۹۸% and strong ROC-AUC values. Descriptive analyses further highlighted consistently high detection rates across sites, with disproportionately elevated standardized rates in smaller and remote communities. Feature importance analysis revealed complementary insights: Random Forest emphasized site-related and demographic attributes, whereas XGBoost highlighted temporal factors such as week and day of sampling. These findings demonstrate the feasibility of applying machine learning to wastewater surveillance, providing proof-of-concept evidence that such approaches can support public health decision-making and inform sustainable water resource management.

Keywords:

Wastewater-based epidemiology , SARS-CoV-۲ , COVID-۱۹ , Machine learning , Random Forest