Test & Attend!

In many low to middle income countries, schools have been closed for more than one year, for underserved students the consequences have been particularly dire. Vaccines, extensive testing and contract tracing have enabled resource-rich countries to reopen schools, but lower income countries lack these resources.

Our solution, Test & Attend, combines breakthroughs in mechanism design and COVID testing to allow schools to reopen safely while making the most of the limited resources on the ground. It was developed in collaboration with experts at the University of Oxford and IPICYT, Mexico.

Every week, Test & Attend screens all students for COVID in small groups before they can attend in-person lessons. A fair scheduling algorithm is used to guarantee equitable access to in-person and online classes for students from all backgrounds. Our solution can provide a blueprint for schools in developing countries all over the world to reopen in a COVID-safe way.

As the SARS-CoV-2 virus continues to sweep across the globe, its devastating effects on children's future are becoming increasingly clear. Lockdown measures designed to combat the COVID-19 pandemic have led to the closure of educational institutions around the world. The impact of these closures on the education and wellbeing of students has been particularly severe in low- and middle-income countries, where many lack crucial resources for remote learning, such as internet access and digital devices. For example, the UN estimates that over 500 million students cannot participate in remote education, and in Mexico, it is projected that the pandemic will lead to about 1.4 million students dropping out of their education altogether. Continued school closures also exacerbate existing social divides. Not only do underprivileged students cope less well with online-teaching substitutes, but many initiatives designed to inspire students from disadvantaged backgrounds to pursue higher education have been suspended, further setting back efforts to improve the equity gap in education.

Our solution will allow many children to return to in-person education while greatly reducing the risk of infections occurring at school. Combining an innovative high-sensitivity pooled testing protocol with optimization methods, we partition students into small groups (clusters) that we test with a single pooled test in order to allow as many students as possible to participate in in-person teaching. By testing clusters on a weekly basis, we ensure that clusters who may infect others participate remotely. The clustering and testing methods are designed to ensure fairness and equal access for underprivileged students, and through continuous evaluation we aim to improve the efficiency of our mechanism over time.

As a pilot study, we propose a workshop to foster skills in AI and high-performance computing for high school students from disadvantaged backgrounds in the state of San Luis Potosí, Mexico. Participants are taught the key scientific concepts behind the strategy that keeps them safe. As part of the curriculum, students are encouraged to improve on the solution with their insights and suggestions. The direct involvement of workshop participants will increase trust and ownership, contribute to learning, and provide valuable feedback for our solution.

The SARS-CoV-2 pandemic has created deep inequalities for students of different means around the world. In low- and middle-income countries, remote teaching has exacerbated existing inequalities. Students with underprivileged backgrounds struggle especially to achieve the desired learning outcomes through online teaching: they may lack laptops or tablets, internet access, or study space. Although recent work tackles the problem of reopening strategies for educational institutions, solutions typically do not take into account the extremely limited testing resources in these communities, nor do they take an active approach in communicating with marginalized communities to allow them to take ownership of the solutions.

Our proposed solution directly involves the students and teachers, by giving them the opportunity to contribute with their unique perspective. From the very beginning of our reopening strategy, students provide information on their particular needs and potential obstacles via an intake survey. The clustering protocol we have designed takes into account multifaceted aspects of fairness, equal access, and individual needs, and is continuously updated in order to ensure that these needs are being met. We have already started engaging with local community groups and teachers through surveys, and the overwhelming response we obtained was that they want schools to reopen, but only when it is safe to do so. For example, many would feel safe to return provided a comprehensive testing strategy was put in place, together with the provision of personal hygiene items such as soap, face masks, and sanitizing cleaning liquids.

The problem we are addressing is closely aligned with several dimensions of the Challenge: our solution is based on hybrid learning where students are not only passive participants, but are contributing to solving the problem of how to reopen schools. The communities we are targeting have been disconnected from teaching during the pandemic. By bringing back in-person teaching, our solution will re-engage these students. Our COVID-19 group testing protocol will ensure that students are tested and safe at school. Working with the You-i-lab in SLP, we are deploying best practices for teaching critical digital skills to the students.

While our novel testing protocol is currently being rolled out across a major university in Mexico, this project introduces novel features such as introducing a highly interactive curriculum teaching digital skills to high school students, and working with the local communities to strengthen our proposed solution. The specific community has not yet been identified, but will be in the municipal area of San Luis Potosi City, SLP, Mexico, where our partner organization You-i-lab is based.

Many public health interventions have a West-centric perspective, a fact that has been clearly identified during the current pandemic. This can be seen, for example around, the issues of vaccination and testing, where schools in developing countries have far fewer resources than do their counterparts in the developed world. Our solution is innovative along two axes: firstly, by approaching the question from a resource allocation perspective we have identified that Western solutions are far from resource-optimal, whereas resource limitations are a central feature in our solution. Secondly, we propose a bottom-up rather than top-down approach to reopening schools, since trust towards authorities is low in many underserved communities.

Over the next few months, our workshop is estimated to serve between 250-500 students. We hope to continue working on this project and scale up to include at least 5000 students within the San Luis Potosi community during the first half of the next academic cycle. We also aim to collaborate with NGOs and expand to other regions in the country, and internationally.

How many infections were prevented due to the testing strategy implemented, compared to running the workshop without a testing strategy? The latter question will be addressed through simulations. 

Has the workshop inspired students to pursue a STEM career, especially among marginalized communities? This question will be evaluated by means of a randomized controlled trial that measures pre and post within-subject change; that is, the perceived change in abilities and aspirations of subjects before and after the workshop.

Volunteers: 7

Our team members are all members of the Mechanism Design for Social Good (MD4SG) Development group, a multi-institutional, international initiative which focuses on applying methodologies from the fields of artificial intelligence, optimization, and game theory to serve under-resourced communities in developing countries. Two of our team members come from Mexico, with close knowledge of and contacts in the region we target in our pilot study. These contacts were essential in setting up a focus group informing the design and goals of our proposed solution, and they provide continued feedback. We also have one team member each from China, Ethiopia, Germany, India, the Netherlands/United Kingdom, and Sweden. Most of our team have worked and lived in LMICs for extended periods of time. They have backgrounds in economics, computer science, and physics, and several have special expertise in education science, development economics, and fairness and equality in mechanism design. They also have previously executed several successful projects in LMICs.

Our team comprises members from various countries across the world, and brings in diverse perspectives. Our team members range from undergraduates to doctoral and postdoctoral researchers, with expertise in a variety of subjects including computer science, economics, and education, providing an excellent opportunity for interdisciplinary collaboration. The team structure is horizontal and everyone’s opinion is welcomed and heard. In addition, we are working closely with the Mexican research community and have made all efforts to directly include them in the research and innovation process. Our goal is to maximize our impact and provide disadvantaged communities equitable access to education by using and combining cutting-edge research from multiple disciplines.

As a Solver, we could scale our Test & Attend solution across many more schools in Mexico beyond our initial pilot. The grant money would be invaluable in developing the scalable software we envision. The monetary benefit would directly serve our target community. Importantly, being selected by Solve would also aid in validating our vision. Our experience working with Mexican policy makers is that this validation, especially from U.S. based institutions, is extremely important in order to take projects off the ground.

A large part of scaling up our solution will be to establish a robust online optimization platform that can be used by many educational institutions in order to reopen safely.

UNICEF, RAISE (https://raise-mit-edu.ezproxyberklee.flo.org/), WHO, PIT Policy Lab (https://www.policylab.tech/), MIT initiative for Responsible AI for Social Empowerment and Education, Social Impact Enterprise Investors. 

These organizations could help us by both suggesting improvements to our solution, and crucially, help implementing at many more communities across the world. Our experience is that it is always best to approach organizations with existing close links to the communities that they are serving, in order to modify the solutions so that they have a strong chance of being effective in the local context.

Yes. Refugees are particularly vulnerable and school closures risk increasing their separation from society even further. Our solutions can be adapted to many refugee communities around the world.

Yes. Our solution addresses directly the issue of equitable classrooms, since school closures across the world have completely cut off low-resourced students from education. With low or no access to digital technologies, these students risk becoming a lost generation of learners, setting back development goals by decades.

Women and girls are particularly vulnerable to extended school closures, and evidence has shown that the number of child marriages have increased drastically during the pandemic. Safeguarding young girls independence and well-being is one of the main motivators for reopening schools as quickly as possible.

Yes, our solution is based on insights from machine learning such as
clustering, optimization, and fairness of algorithms, and will be used
to greatly amplify the reach and efficiency of extremely limited COVID
resources such as tests. These methods could allow many children from
disadvantaged backgrounds return to in-person education.

Yes, our solution is based on insights from machine learning such as

clustering, optimization, and fairness of algorithms, and will be used
to greatly amplify the reach and efficiency of extremely limited COVID
resources such as tests. These methods could allow many children from
disadvantaged backgrounds return to in-person education.

Yes, our solution is based on insights from machine learning such as

clustering, optimization, and fairness of algorithms, and will be used
to greatly amplify the reach and efficiency of extremely limited COVID
resources such as tests. These methods could allow many children from
disadvantaged backgrounds return to in-person education.