Modelling outbreak investigation following environmental surveillance alerts

Michaelmas 2026 project

Joseph Hicks

Summary

Earlier pathogen detection creates an opportunity for faster investigation, but the benefit depends on what happens after an alert. Environmental surveillance, including wastewater testing, can identify pathogen signals before infections are recognised clinically. However, a signal does not necessarily establish that people are even infected. Uncertainty and limited investigation capacity can delay confirmation, reducing the advantage of early detection.

 

One possible response is to deploy an investigation team to collect samples from people in the affected community. This could establish human infection before routine clinical surveillance, but deployment and testing take time, and sampling may fail to identify infected people even when infections are present.

 

This project asks: when is it worth deploying an outbreak investigation team for community testing following an environmental surveillance alert, rather than waiting for cases to be identified through routine clinical surveillance?

What you will investigate

Together, we will select one jurisdiction and setting from a shortlist with accessible public documentation. You will map the pathway from an environmental alert to an initial investigation, identifying who receives the information, who makes deployment decisions, and what must happen before samples can be collected and tested. This focused exercise will inform the model rather than comprehensively assess the jurisdiction’s preparedness.

 

You will then build a simple mathematical model in R comparing routine clinical surveillance with and without an additional investigation triggered by an environmental alert. The model will also explore how the comparison changes under different assumptions about mobilisation delays, probability of detecting infection through sampling, laboratory turnaround, and clinical recognition. Outcomes will include the probability of earlier confirmation, time gained, and the number of people tested. We will also account for investigations that do not confirm infection during the study period.

What you will produce and learn

The minimum outputs will be a documented pathway map, reproducible R code, and a short report explaining findings and limitations. We will spend the first two weeks narrowing the question, reviewing evidence, and agreeing on model structure. We will then develop and explore the model before reserving the final two weeks for finalising visualisations and writing. 

You will help shape the scenarios and learn how to turn a broad public health question into a tractable study, represent a real process mathematically, and distinguish conclusions supported by evidence from those dependent on assumptions. 

Suggested background reading:

Impact case

The findings could help surveillance programme designers identify the investigation capacity needed to make use of early alerts. They could also inform public health discussions about deployment by quantifying potential gains in confirmation time alongside investigation effort. Building on existing examples of surveillance-triggered testing, the project will offer a transparent framework for examining operational trade-offs and identifying evidence gaps.

Student background

This project suits students interested in infectious disease, pandemic preparedness, and quantitative policy research. Prior biology or epidemiology knowledge is not required. Complementary strengths in programming, critical reading, and communication are welcome. At least half the team members should be able to write and debug R code independently while the other team members should be willing to develop their coding skills. 

  • Interest in outbreak preparedness and surveillance.

  • Basic probability and statistics; R or Python experience desirable.

  • Willingness to examine assumptions and interpret uncertainty.

  • Ability to collaborate and communicate findings clearly.