In the last week of March 2021, Carol lost her colleague to a brief illness. Her co-worker, a 28-year old IT assistant, fell ill with a fever following a quick trip to his village. Two days later, he was gone and three days after that, Carol and several other members of her team were packed into a tent listening as family members paid tribute to one of their own. Carol’s trip back to the city was uneventful save for her seatmate who seemed to be nursing a bad cold. Aside from removing her mask to take a few sips of soda, she thought there was nothing to worry about as she was wearing it the rest of the time. However, five days later, as Carol sat in her house nursing a flu, she realized she could neither taste nor smell the orange she was eating and she wondered aloud, “Could this be what I think it is?”

Across the country, Mrs. Bangura was on a phone call to her son. Two days prior, the government declared a pandemic in the country and all movement was restricted. Her son, who had traveled to the city for a meeting, was now stuck there. While on one of their phone calls, she informed him of his father’s worsening flu. “Ever since he returned from that funeral last week, it has been one thing after another,” she said. Her husband had escorted his friend upcountry to bury his son, a bright young man whose sudden death had shocked the community. Upon his return, however, Mr. Bangura had been feeling ‘off’ but his wife’s pleas to see a doctor had fallen on deaf ears, so she hoped that her son could talk some sense into his father. Further increasing her worry was the news filtering in from the media as apparently, people who suffer from chronic diseases were more at risk of contracting this new virus. Her husband had been hypertensive for over 15 years and she shuddered to think of what would happen to him if he got that “whatchamacallit”.

Scenes like Carol’s and Mrs. Bangura’s were replicated across the country as citizens battled the COVID-19 virus at a personal level and as a collective. Since it first emerged in December 2019, governments across the world put in place various control measures to curb the spread of the disease. Simultaneously, medical and public health research institutions were scrambling to know more about the disease—how did it come to be, how long does it stay in the body, when will the next wave be, and how many variants are there? Owing to the uncertainty surrounding COVID-19, it was and still is imperative that governments and non-state actors work together to develop systems that provide proper epidemiological planning and management, especially where health systems are weakest.

It is for this reason that the African Population and Health Research Center (APHRC) together with Malawi Polytechnic, the ALPHA network of the London School of Hygiene & Tropical Medicine (LSHTM), the South African Population Research Infrastructure Network (SAPRIN), and the Committee on Data of the International Science Council (CODATA) collaborated to establish a data hub for COVID-19 data in Kenya and Malawi. This study, a component of SIDA-IDRC’s Global South Artificial Intelligence (AI) for COVID-19 (AI4COVID) program, aimed at ensuring that data goes beyond counting the numbers to address the unintended risks and consequences of communities that are excluded or under-represented. This type of layered data can provide government institutions with a full picture of the pandemic for informed and holistic policymaking.

As governments continue implementing programs to bring the spread of the virus to a halt, they must also develop their interventions in ways that prevent, reduce or alleviate the layered negative impact on at-risk populations. At the same time, they can lean on technology such as AI and machine learning for more accurate disease mapping, including timings, locations, and groupings. In this way, they can preempt when the next wave of the pandemic will be, which populations are most likely to be affected, and perhaps even the severity of the cases based on previous trends. As part of its contribution to the fight against COVID-19, the INSPIRE network set up the Platform for Evaluation and Analysis of COVID-19 Harmonized data (PEACH) in 2020 to develop a template for sharing, harmonizing and using health data from population cohorts in Kenya and Malawi. The Global South AI4COVID program provided a way to test the Observational Health Data Sciences and Informatics (OHDSI) common data model we adopted to see whether we could combine data and answer policy-relevant questions about the COVID-19 pandemic. “We discovered this is a lot more difficult than we thought for several reasons. Still, the experience of working with partners across Africa has enabled us to look at how AI can be used in public health research in Africa,” said Professor Jim Todd of the London School of Hygiene & Tropical Medicine (LSHTM) who was a co-investigator on the study.

Although the worst of the pandemic has now passed, the threat still looms in the shadows, with many governments put on the spot, and forced into action within a limited policy space and with already limited resources, particularly in low and middle-income countries (LMICs). Ministries of Health across the continent are continually running vaccination campaigns and proof of COVID-19 vaccination is mandatory for cross-border travel across many African countries. 

As the study comes to a close, the study team has learned several lessons along the way, such as about access to data and the day-to-day management of such an initiative across borders. “Owing to the levels of management involved in managing the disease at the country-level, obtaining data from official sources proved a challenge. Also, recruiting and training data managers to work on the platform took more time than we had initially anticipated,” said Dr. Sylvia Muyingo, Associate Research Scientist at the African Population and Health Research Center (APHRC).  These challenges were overcome through data from health and demographic surveillance (HDSS) sites and smaller datasets from similar studies. Eventually, the government of Malawi agreed for MUBAS to temporarily host the data servers with a plan to transfer them to the Ministry of Health in the near future.

Dr. Muyingo adds, “currently, most data systems in Africa are fragmented even though the integration of national and sub-national data systems is still needed to drive continued response and recovery from the pandemic. At the same time, while many African countries support inclusive AI solutions, the guidelines for these remain limited. Moving forward, countries will require the use of various sources of data to inform public health decision-making for effective pandemic response.”

Furthermore, “the PEACH program made us realize that although technology is important, we will not achieve anything without addressing human concerns, such as data safety, confidentiality, how AI impacts gender rights, and the ethics behind the analyses undertaken,” states Professor Todd of the London School of Hygiene & Tropical Medicine (LSHTM).