Introduction to WBE case estimation: A practical toolset for public health practitioners

Area of Science:

• Environmental science
• Epidemiology
• Public health

Background:

• Wastewater data offers insights into disease dynamics like incidence and prevalence.
• Analyzing wastewater data often requires advanced expertise beyond entry-level epidemiologists.
• Predictive modeling from wastewater can support public health decision-making.

Purpose of the Study:

• To predict COVID-19 cases using wastewater data.
• To develop a customizable, low-resource toolkit for local public health jurisdictions.
• To evaluate different modeling approaches for wastewater-based epidemiology.

Main Methods:

• Utilized literature-derived viral shedding and fecal production rates.
• Incorporated municipality flow rate data.
• Analyzed wastewater-derived viral concentrations to predict COVID-19 infections.
• Evaluated four distinct predictive models, including a gradient boosted decision tree and a deterministic calculation.

Main Results:

• The probabilistic gradient boosted decision tree model showed the highest accuracy (MAE 0.09%-0.25%).
• A simpler deterministic calculation model performed nearly as well (MAE 1.2%).
• A probabilistic method yielded the least accurate results (MAE 3.6%).

Conclusions:

• Public health jurisdictions can use various models to predict infectious disease cases from wastewater.
• Model selection should consider data availability, technical expertise, and community-specific needs.
• Tailoring models to local contexts is crucial for effective wastewater-based epidemiology.