The Power of Collaboration, Artificial Intelligence and Big Data in the fight against COVID-19 in Africa

Authors: The ACADIC Research Team

Thursday, 30 September 2021

With the scarcity of COVID-19 vaccines in many African countries and new variants on the rise, real-time delivery of reliable and comprehensive information is crucial for guiding government policies and practices. Supported by Canada’s International Development Center (IDRC) and the Swedish International Development Agency (Sida) as part of the Global South AI4COVID program, the Africa-Canada Artificial Intelligence and Data  Innovation Consortium (known as ACADIC) is providing locally-nuanced data insights to inform public health decision making, as well as vaccination roll-out strategies.  With a view to scaling up approaches across the continent, the team is leveraging big data and AI-based techniques in nine African countries, namely Botswana, Cameroon, Eswatini, Mozambique, Namibia, Nigeria, Rwanda, South Africa and Zimbabwe.

Below we discuss some of the main components of our research:

Obtaining locally relevant data to train AI-based algorithms can inform evidence-based, data-driven policy making on public health. However, adopting such an approach in Africa can be challenging. For instance, whilst community leaders are usually aware of what is happening in their areas during an outbreak, the pathways for communication with experts who can help address the problem are limited. In light of this challenge, ACADIC brings together local authorities, public health authorities, politicians and researchers for discourse and action, particularly to secure access to locally relevant data and inform model formulation. Mapping the issues at hand with community leaders, the research team then explores and identifies datasets that might be leveraged. Community-based participatory action is central to this approach, which allows the team to develop models that are tailored and contextualised to address the needs and challenges that are specific to each community. 

Providing governments with an accurate portrait of the epidemic at the local, state, and national levels is also crucial. In communities across Africa, there are stigmas attached to being infected with the COVID-19 virus. Moreover,  self-medication and the use of complementary or alternative medicine is commonplace due to social, economic and psychological reasons. These factors contribute to under-reporting, which affect governments’ ability to have a clear sense of the nature of the epidemic. To address this issue, ACADIC employs an augmented susceptible-exposed-infectious-recovered modelling approach to estimate the under-reported values and determine the truce case count. The research team is using the results from this modelling to train and refine an AI algorithm that has been developed to provide alerts on when it’s safe for communities to transition back from full lockdown to business as usual. 

Using AI-powered data visualizations, governments can optimize their response strategies. Working with local communities, ACADIC has been harnessing AI and Big Data  techniques to develop targeted vaccination roll-out strategies, particularly focused on the at-risk population. Deploying and administering vaccines across large populations requires dedicated planning and resources, and by using AI approaches to identify hotspots and the population at risk,  governments can optimize resource allocation. The research team has developed a COVID-19 dashboard that automates data visualizations to show how the outbreak is evolving. As an accountability tool, this helps to support timely exchange of information between policy makers and the public. These visualizations can help citizens to make informed decisions, and provide insights to tackle mis- and dis-information about covid-19 prevention, treatment and vaccines.

Providing near-term forecasts (nowcasting) of labour market flows is a priority for governments during the pandemic as they attempt to balance decisions on public health with the economy. The official unemployment rates across African countries tend to be released infrequently with substantial delay. This however has hampered policy makers in accessing timely information given the speed of the economic decline at the onset of COVID-19 outbreak. To address this concern, our research team  immediately began employing artificial intelligence using proxy indicators to provide insights on the unemployment rates in countries across Africa. As each country begins to track progress of its socio-economic recovery, being able to forecast the timing and extent of improvements in the labour market outcomes will also be useful for policymakers. 

ACADIC is an interdisciplinary team of data scientists, epidemiologists, physicists, mathematicians, software engineers, as well as disaster and emergency management, clinical public health, citizen science, and community engagement experts, coming from the following organizations: 

1) The Council for Scientific and Industrial Research (CSIR) - South Africa

2) The South African Medical Research Council (MRC) - South Africa

3) University of the Witwatersrand - South Africa

4) African Institute of Mathematical Science (AIMS) - Rwanda

5) University of Ibadan - Nigeria

6) University of Buea - Cameroon

7) Ministry Public Health - Cameroon

8) African University of Science & Technology - Nigeria

9) iThemba LABS - South Africa

10) Rhodes University - South Africa

11) University of Botswana - Botswana

12) Ministry of Higher Education, Technology and Innovation- Namibia

13) Namibia University of Science and Technology - Namibia

14) Universidade Eduardo Mondlane - Mozambique

15) National University of Science and Technology - Zimbabwe

16) University of Eswatini - Eswatini

17) Laboratory for Industrial and Applied Mathematics (LIAM) -York University 

18) The Dahdaleh Institute for Global Health Research - York University

19) The Advanced Disaster, Emergency and Rapid Response Program - York University

To learn more about our project, check out our website and follow us on Twitter @theACADIC

Tags: #Community-based, #HarnessingAI, #BigData, #AI-Powered