Bringing big data science to Africa
Big data and artificial intelligence techniques, such as machine learning, are changing the very nature of science, making it possible for scientists to glean deep insights from massive volumes of data. But an important question remains: How can these new techniques help Africans?
A special event taking place this week in Akure, Nigeria, aims to not just answer this question, but help bring these powerful tools to a new generation of African researchers. The event is bringing together over 70 new and established researchers in this field from 14 different countries – 8 of them African. It is organised by TWAS and the TWAS Young Affiliates Network (TYAN), and co-organised and supported by the Elsevier Foundation, a long-time TWAS partner dedicated to diversity and inclusion in science, technology and health, as well as advancing research in developing countries.
Bolanle Ojokoh, a data scientist who serves on the Executive Committee of the TWAS Young Affiliates Network (TYAN), said that knowledge about artificial intelligence and big data is not yet readily available in Africa.
“So we expect this event to provide a platform,” Ojokoh said. “We have people who can exchange ideas. We can interact, and it will help to gain knowledge from one another. At the end of the day this will build capacity for all of us as researchers in this part of the world.”
“Researchers in Africa can contribute greatly to the data science revolution, which is rapidly accelerating contributions to the U.N. Sustainable Development Goals – but it is essential that they are part of this conversation and global research journey,” said Elsevier Foundation Director Ylann Schemm. “The Elsevier Foundation is proud to support the TYAN/TWAS symposium in big data, analytics, and machine intelligence co-hosted by the Nigerian Young Academy.”
The event is called the 4th TYAN International Thematic Workshop and 1st African Symposium on Big Data, Analytics and Machine Intelligence for Financial, Health and Environmental Inclusion in Developing Countries. It begins with a symposium, including talks by several experts in the field, and concludes with a workshop in which African researchers will receive hands-on training in big data and machine learning techniques.
Ojokoh, who helped organise the event, said that African nations typically lack not only the equipment with which to explore the use of these data-heavy techniques, but access to experts who can teach practical applications of these tehcniques. At the workshop, they will learn how to use these data-analytics tools, and many of the speakers will demonstrate how they work.
For example, Congolese population geneticist Emile Chimusa is currently a researcher at the Institute of Infectious Disease and Molecular Medicine in Cape Town, South Africa. He is one of the event’s co-organizers, and called the workshop critical to Africa.
“The idea is to bring most of the experts around the world and to discuss in a kind of panel discussion, what is the challenge and how to go about that challenge,” Chimusa said.
Chimusa uses bioinformatics to study how resilient or susceptible African populations are to diseases, such as tuberculosis, HIV and breast cancer. He says machine learning – a category of artificial intelligence that automates big data analysis to identify patterns and shape models with little human help – is particularly important to work such as his.
The reason for this? Within the African continent there is a particularly high level of genetic diversity. But existing machine learning techniques are designed for populations in the West that are genetically “bottlenecked”, and thus have less diversity. So it’s not reliable to use these programmes to study African disease resilience. This means Africans need to understand how these machine learning programmes are composed so they can make their own.
“When African populations start to generate big data and that data has been generated based on external funding and sent and analyzed overseas, and however most of the machine learning to pinpoint the location of disease vulnerability in the DNA cannot work because the African population is more diverse,” said Chimusa. “So there’s a need to develop more machine learning tailored for African genetics characteristics and for populations that are highly diverse or highly mixed.”