Applying CoronaSurveys to public health emergencies

Building on the CoronaSurveys experience, we propose the use of indirect surveys to detect and monitor public health emergencies, and to instrument measures to fight them

Antonio Fernández Anta

Research Professor

IMDEA Networks Institute

The CoronaSurveys project started in March 2020 with the goal of deploying indirect surveys to estimate the number of cases of COVID-19, since most countries had limited testing capacity at that time. Since then, these surveys have been made available in all countries of the world, and are now being  used to monitor other aspects of the pandemic, including testing, vaccination, and incidence of long-Covid. This is especially useful in countries with limited resources. Additionally, the CoronaSurveys team has developed techniques that leverage other data sources to forecast the pandemic evolution and to recommend interventions.

The techniques developed in the CoronaSurveys project for the COVID-19 pandemic can be used in the future to detect and monitor other health emergencies, and to respond to them with informed measures.

The main objective of this project is to develop a powerful tool in the fight against the pandemics. In that sense, the main target audience are the decision makers that can use the alarms, information, estimates, and forecasts that are produced, and can evaluate the interventions the project will propose. We are especially interested in reaching decision makers in geographical areas in which the infrastructure to collect data is limited (like LMIC). We are reaching out to governments and health officials, in order to identify their needs. For instance, we are directly involved with a COVID-19 Committee of the Regional Government of Madrid, and will start producing periodic reports to them in the next few weeks.

In addition, we are also targeting the general public, and for that reason we make all our data and results public. The motivation for this is two-fold: (i) the information we provide can be useful, especially if official sources of information are unreliable, and (ii) providing data and results helps us motivate the general public to participate in the survey and spread the word about the project.

The solution we propose provides public good in several dimensions. The most direct benefit is the information (estimates) about the pandemic, which is being produced and made publicly available via the project website and GitHub repository. Another direct good is the collection of survey responses that are made public also via the GutHub repository, with which anyone can perform analyses. The project is also open-sourcing the software generated in the project, for anyone to use. Finally, and possibly most importantly, we are publishing scientific papers describing the open surveys system and related data analysis  techniques, including some that are novel and address the specific challenges of online data collection.  We are making all our publications freely available (through Open Access publication and/or posting on arXiv/medrXiv).

The solution will have an impact in the evolution of the pandemic by providing data about it that is not readily available. Having reliable data is very important for making informed decisions. We have shown that our estimates of cases in Spain in the Spring of 2020 are more reliable than the official numbers. Case estimation is now very reliable in many countries, but there are still LMIC countries that may benefit from our data. Moreover, there are aspects of the pandemic (such as the incidence of long covid) that are not known in most countries, and that we can help estimate.

Additionally, the algorithms we are developing to integrate our data with other data available (like the Symptom Surveys) will provide a much more complete (daily) picture of the pandemic, and will suggest measures to decision makers whose effectiveness is rooted in all this available data.

We will, in particular, provide the Government of Madrid with periodic reports of the regional situation that will complement the data they collect, and will propose intervention plans to them with estimated effects in health and economic parameters.

The solution is already deployed and providing data for all countries of the world. However, the participation in the survey has been very limited geographically, and mostly centered in Western European countries. If funding is available, we will launch campaigns in a few LIMC representative countries, in order to reach a sufficiently high level of  participation. These campaigns will use any available means, like social networks, phone surveys, and mass media. In order to maximize the impact we have local contacts in several of these countries (like Ecuador, Brazil or Nigeria) and will make an effort to have more.

These actions to be done during the first year in the context of the COVID-19 pandemic will be extended beyond it (to 3 years and more) but to monitor other issues, like detecting early the outbreak of an epidemic, evaluate poverty and hunger levels periodically, etc.

The main metric we would like to use to measure the impact of the solution in the pandemic is lives saved. Unfortunately, that is hard to measure. Instead, we will be measuring other indicators, such as participation in the survey, traffic in the website and GitHub repositories, regions and countries with which we have direct collaboration, scientific papers published, citations to our papers, software products released, etc. These indicators will provide a complete picture of the impact and its evolution over time.

Currently we see two main barriers for achieving the desired impact with the solution. The first barrier is financial. The project has had very limited funding so far, which has not allowed us to deploy large-scale paid campaigns to raise participation in the survey or to  have researchers working full-time in the project (all team members have been working part-time on a voluntary basis). The second barrier is accessibility to decision makers and authorities, in order to help them make informed decisions. (They could also help raise participation locally.) We have been very actively recruiting and contacting people from multiple countries, with limited success in reaching the authorities, and there are still many countries that could benefit from our solution and where we currently have no contact with the health authorities. This challenge and the partnership with Facebook will make a great presentation card to increase the effectiveness of our contact attempts.

IMDEA Networks Institute

Madrid, Spain

We are applying to the Trinity Challenge because we want the CoronaSurveys project to be useful for the COVID-19 pandemic and beyond. For that, we need funding to fully develop the system proposed, support to make it visible to stakeholders and decision makers, and the partnership of a Challenge member. 

On the one hand, part of the funding will be used to maintain the computational infrastructure, and to enroll researchers and developers that can work in the project full time. On the other hand, the funding will be used to increase participation in the surveys, in order to have a steady flow of data (responses). 

The visibility that the Trinity Challenge will provide to the project will be very important to increase the trust stakeholders and decision makers will have in the CoronaSurveys project and team.

Finally, as described below, the partnership with Facebook will provide increased participation and access to important data

We would like to partner with Facebook, following several months of interactions. Facebook invited the CoronaSurveys team to participate in the COVID-19 Symptom Data Challenge, which introduced us to the COVID-19 Symptom Surveys data being collected by CMU and U Maryland with their help. Since then, we have obtained access to the detailed data of the surveys, and are using it in combination with our survey data. However, we hope that the collaboration in the context of the Trinity Challenge will go beyond the COVID-19 Symptom Surveys. Tamer Farag, Strategic Partnerships Manager for Health at Facebook, has confirmed the willingness of this partnership.

Facebook could support the solution by promoting the survey. Facebook has access to a large population of users that could be easily reached and invited to fill the survey. In addition, we would like to work closely with Facebook research teams in order to explore how to use and improve the techniques and algorithms used in the project, and the network scale-up method in particular. For instance, one aspect we would like to explore is how Facebook data can help identify if these methods present biases, and if so how they can be removed.