- Sputnik Africa, 1920
Global South Pole
Global South Pole is more than just a podcast. It’s a platform dedicated to challenging the mainstream narratives and amplifying the voices of the overlooked communities. It’s time to rewrite the maps to plant the Global South at the center.

Why African Data Matters in Future of Medical AI

Why African Data Matters in Future of Medical AI
Subscribe
Artificial intelligence is used to study diseases, predict outbreaks, and support medical research, but its effectiveness depends on the data used to train it. With African populations underrepresented in global genomic datasets, questions remain over how accurately AI systems reflect the continent’s genetic diversity and public health realities.
The data gap could have consequences for disease diagnosis, outbreak prediction, treatment, and the detection of antimicrobial resistance. In one example discussed during the program, a model for predicting ciprofloxacin resistance reportedly achieved about 75% accuracy in England but around 50% when applied to African data.
Global South Pole spoke to Dr. Oladipo Elijah Kolawole, an associate professor of infectious diseases, molecular biology, bioinformatics, and genomics at Adeleke University in southwestern Nigeria and founder and principal investigator of the Helix Biogen Institute in Ogbomoso.
Kolawole said AI models trained predominantly on data from other populations may not adequately capture Africa’s biological diversity, potentially limiting their usefulness when applied to African populations. He argued that the continent needs to generate more of its own genomic data and use it to train models that are more specific to African populations and health conditions.

"If you look at most of the databases that we are working with, like you spoke about, we have little or few disasters of Africans that are dead. And because we have the highest diversified genetic components in the world [...] it means that prediction in terms of diagnosis will not be accurate. It means the treatments in terms of accuracy will not be accurate. We need to go back as an African, not as an other entity, to develop our own personalized model that we're going to use—an African data model. We're going to use the African data set to train the model so that we can have something that is specific to us and sensitive to us. Even when it comes to diagnostics, we need our own data to train, to develop, and to model for us to have a better position," Dr. Kolawole stressed.

To listen to the whole discussion, tune in to the Global South Pole podcast, brought to you by Sputnik Africa.

In addition to the website, you can also catch our episodes on Telegram.
► You can also listen to our podcast on Apple Podcasts, Spotify, Deezer, Castbox, Pocket Casts, Afripods, Podcast Addict, Overcast, and Mave Stream.
► Check out all the episodes of Global South Pole
Newsfeed
0