Detecting COVID-19 infection hotspots in England using large-scale self-reported data from a mobile application: a...

Area of Science:

• Epidemiology
• Public Health
• Digital Health

Background:

• Tracking COVID-19 spread is crucial for targeted interventions without national restrictions.
• Local-level disease surveillance is essential for effective public health policy.

Purpose of the Study:

• To model COVID-19 incidence and prevalence using self-reported data from a mobile application.
• To estimate the effective reproduction number (R(t)) at a granular geographical level.
• To validate the app-based model against established surveillance data sources.

Main Methods:

• Prospective, observational study using longitudinal data from the COVID Symptom Study app (March-September 2020).
• Utilized self-reported symptoms and RT-PCR test results for incidence and prevalence estimation.
• Employed logistic regression, Poisson process, and Markov Chain Monte-Carlo for modeling and R(t) estimation.
• Validated findings with Office for National Statistics (ONS) and REACT-1 study data.

Main Results:

• Over 2.8 million users contributed data, with 169,682 RT-PCR tests recorded.
• National estimates of incidence and prevalence aligned with ONS and REACT-1 findings.
• On September 28, 2020, estimated incidence was 15,841 daily cases, prevalence 0.53%, and R(t) 1.17.
• Identified 15 of the 20 highest incidence regions using geographically granular data.

Conclusions:

• Self-reported mobile app data provides an agile resource for pandemic surveillance.
• The methodology can detect rapid case increases in areas with limited government testing.
• App-based surveillance complements traditional methods for informing policymakers during public health crises.