A multi-city COVID-19 forecasting model utilizing wastewater-based epidemiology

Area of Science:

• Environmental Science
• Epidemiology
• Public Health

Background:

• COVID-19 forecasting models have shown limitations, particularly in performance during critical outbreak periods.
• Reliable data inputs and alternative forecasting targets are crucial for effective public health decision-making.
• Wastewater-based epidemiology (WBE) offers a promising, more consistent metric for tracking disease transmission compared to reported cases.

Purpose of the Study:

• To develop and validate a multi-city, wastewater-based forecasting model for predicting COVID-19 hospitalizations.
• To introduce novel categorization methods for assessing healthcare system burden and hospitalization trends.
• To evaluate the impact of WBE data on forecasting accuracy, especially during outbreak change points.

Main Methods:

• A Generalized Additive Model (GAM) was developed using hospitalization and COVID-19 WBE data from six US cities.
• Two categorization types were generated: Hospitalization Capacity Risk (HCR) and Hospitalization Rate Trend (HRT).
• Probabilistic forecasts were created for 1, 2, and 3-week windows, with a new methodology for assessing performance at outbreak change points.

Main Results:

• The inclusion of wastewater data significantly improved the forecasting model's performance.
• The model successfully generated probabilistic forecasts for hospitalization risk and trend categories across six cities over 20 months.
• A novel method for evaluating model performance at critical outbreak change points was proposed and applied.

Conclusions:

• Wastewater-based epidemiology provides a valuable and reliable data source for COVID-19 forecasting.
• The developed categorical forecasting model offers a new tool for decision-makers to predict hospital capacity risk and disease trends.
• This approach enhances public health preparedness and response strategies for infectious diseases.