What is it about?

During the COVID-19 pandemic, exposure notification apps were developed to scale up manual contact tracing. They automated notifying contacts of infection using information on proximity of mobile devices. Here we present the next step beyond these apps, network data assimilation (DA). Network DA uses this data more effectively, by aggregating it with a model of disease transmission to produce individual user-level risk assessments. We inherit our methods from the way satellite data streams are combined with models of the Earths atmosphere to produce local weather forecasts. Supported by extensive simulations studies of New York City (NYC), we show that network DA predictions track outbreaks more accurately and reliably than exposure notification apps. We also show how individual risk information can be used to target intervention policy, achieving epidemic control by isolating only a small targeted fraction of the population at any given time.

Featured Image

Why is it important?

Current means to prevent spread of COVID-19 (and other infectious diseases) involves sweeping population-wide mandates: quarantines, social distancing and lockdowns. Though they can be effective, such mandates come at a huge cost to economic and social environments. Our findings demonstrate that by leveraging currently available data and models, one can significantly reduce this disruption (isolating only 5-10% population at any time) while simultaneously reducing deaths (by over 50%) compared with blanket interventions. Moreover, network DA is general. It can use a multitude of data sources, and can be applied to other populations and transmissible diseases, giving it great future potential towards effective control of other transmissible disease outbreaks.


I have long taken for granted the importance of having access to local weather forecasts at my fingertips. Yet such information is influential over many aspects of my life; what I will wear that day, the modes of transport I take, and how I might act safely. Thinking back to March 2020, many of us were subjected to an environment without available prediction; here we felt the huge insecurities and the lasting impacts this had on our lives. Big unknowns, big consequences, and sweeping policies. I believe this article takes an exciting step towards a future, when, during a crisis, I can simply check my own personal risk forecast in the morning, and make confident decisions on how to act safely that day. Our evidence suggests that this really is within reach, and reduced loss of life, at reduced cost, can possibly be gained with currently available data, and current models of disease transmission.

Oliver Dunbar
California Institute of Technology

Read the Original

This page is a summary of: Epidemic management and control through risk-dependent individual contact interventions, PLoS Computational Biology, June 2022, PLOS, DOI: 10.1371/journal.pcbi.1010171.
You can read the full text:

Open access logo


The following have contributed to this page