The researchers’ focus has been on childhood infections in East Africa and Southeast Asia. For more than 15 years, they have worked with hospitals, health providers, planners, economists, engineers and policy makers in Uganda on predictive mapping of childhood infections and birth defects.
In early 2020, they began discussing ways to use their skills to contribute to the fight against SARS-CoV-2, the virus causing covid-19. They relied on a wide network of experts, organizations, and local authorities. They helped tailor the approach to generate forecasts for the following week’s numbers of covid-19 cases.
The authors created a set of visualization tools and graphs that were very easy to interpret. They built on a proven method that was already in use in the UK and several European countries, adapting it to the entire African continent.
The scientists used a relatively simple model to base the code on. They incorporated the daily cases from each country, as well as the Human Development Index characterization for each nation, population, the stringency of social measures to control infection, and meteorological data. They took into consideration cases in neighboring countries, as well as whether the country is landlocked.
The model used the numbers of cases reported up to and including the previous week, to predict the numbers of cases the following week.
The model included previous weather observations of temperature, precipitation, and humidity. Some of these environmental factors have been implicated in covid-19 transmission out of Africa.
The researchers took into account the cross-border flow of goods and trade, since for landlocked African countries, this is their lifeline for goods and services.
It is difficult to get population mobility data for Africa, so cases in all other countries were used to test whether those cases were associated with the number of next week’s infections in nearby countries. This proved to be very effective.
In addition, they considered the economy of different countries, as well as the evolution of government restriction policies in real time, such as confinements and border closures.
The predictions were much more accurate than expected. The authors employed a rigorous way of scoring the forecasts for each country. Only for a few countries were these predictions inaccurate (Burundi, Cameroon, Somalia, and Botswana).
You can see the forecasts for each country here.
