Wastewater-based epidemiology for COVID-19 using dynamic artificial neural networks

Area of Science:

• Environmental microbiology
• Epidemiology
• Artificial intelligence

Background:

• Global vaccination efforts have reduced COVID-19 deaths, but high SARS-CoV-2 circulation persists, necessitating advanced monitoring tools.
• Wastewater-based epidemiology offers a powerful, non-invasive method for public health administrations to assess regional disease risk.
• Accurate prediction of COVID-19 hospitalizations is crucial for effective resource allocation and public health management.

Purpose of the Study:

• To develop and validate a dynamic artificial neural network (DANN) model for predicting COVID-19 hospitalizations.
• To integrate wastewater surveillance data (SARS-CoV-2 N1 gene concentration) with vaccination coverage and historical hospitalization data.
• To assess the model's predictive accuracy and its utility in assigning risk levels for regional health management.

Main Methods:

• A dynamic artificial neural network (DANN) was developed using wastewater SARS-CoV-2 N1 gene concentrations, vaccination coverage, and past hospitalization data.
• The model incorporated both instantaneous values and historical trends of the input variables.
• Two distinct study periods (May 2021–September 2022 and September 2022–July 2023) were utilized for model training and validation.

Main Results:

• The DANN model accurately predicted COVID-19 hospitalizations, with strong correlation observed between N1 gene concentrations and hospitalizations (r=0.43, p<0.05) during the first period.
• The model demonstrated accurate forecasting for 1-day and 5-day ahead predictions.
• Retraining the model for the second period maintained accuracy despite lower hospitalization numbers, and risk level assignments showed high agreement (95% and 93%) with health authorities' reports.

Conclusions:

• The developed DANN model shows significant potential for predicting COVID-19 hospitalizations at a regional scale using wastewater epidemiology.
• This approach provides a valuable tool for public health authorities to support COVID-19 risk management and decision-making.
• Wastewater-based epidemiology, integrated with AI, offers a robust strategy for ongoing public health surveillance.