# Community‐level wastewater surveillance with machine learning methods to assess underreporting of COVID‐19 case counts

## TL;DR

This paper uses wastewater data and machine learning to detect underreported COVID-19 cases in communities.

## Contribution

The novel contribution is a high-accuracy machine learning model that identifies underreporting of cases after testing mandates ended.

## Key findings

- A prediction model with high sensitivity and specificity was developed for tracking community-level COVID-19 prevalence.
- The model provides evidence of significant underreporting of cases after testing mandates were lifted.
- Wastewater RNA data from sewage manholes serve as reliable biomarkers for monitoring infections.

## Abstract

COVID‐19 remains an ongoing threat to public health, and reliable, continuous disease monitoring programs are essential for preventing future surges of infection. However, without mandated COVID‐19 testing, accurate data of confirmed cases are unavailable. Instead, COVID‐19 viruses may be tracked via wastewater samples from sewage manholes in areas of high social connectivity, where captured viral RNA data are biomarkers useful for monitoring and predicting community‐level COVID‐19 prevalence through machine learning techniques. We construct a prediction model of high sensitivity and specificity to provide evidence of significant underreporting of COVID‐19 cases for the time period following the lifting of testing mandates.