Advanced data science approach can identify new diabetes risk markers in African populations
- Beth Amato, Wits Faculty of Health Sciences
This could lead to more effective screening tailored to underrepresented groups in genetic research.
Data scientists at the Sydney Brenner Institute for Molecular Bioscience (SBIMB) at Wits University used innovative analytical methods to identify new risk factors associated with Type 2 diabetes in African populations.
Currently, the conventional approach to diabetes screening relies on research primarily conducted on people of European descent.
Why African data matters
According to a recent study in Nature Communications titled "A Stratification Method for Identifying Subgroups at High-Risk for Type 2 Diabetes in sub-Saharan Africa”, co-authored by SBIMB postdoctoral researcher Dr Kayode Adetunji, type 2 diabetes has typically been associated with high-income and Western lifestyles. With increasing urbanisation in Africa, we are moving toward more sedentary lifestyles, poor dietary choices and widening girths, all of which contribute to increased disease risk, with a projection that type 2 diabetes could reach a high of 853 million affected individuals worldwide by 2050.
“Sub-Saharan Africa currently only represents 4% of global cases. But the region faces the steepest increase, at 142% by 2050, signalling an emerging public health crisis. What’s more, the true burden may be underrecognised in African contexts due to limited screening programmes and diagnostic resources,” explains Adetunji.
The study, published in Nature Communications, drew on data from multiple African sites, including those in Mpumalanga, South Africa, East Africa and West Africa, and applied a multidimensional scanning method and optimisation-based algorithms to identify subgroups of individuals with common risk profiles for type 2 diabetes.
The study challenges the reliance on screening guidelines based on populations from the Global North, refines screening criteria, and could reduce the number of people whose risk goes undetected.
"We're not saying that the current screening frameworks are bad," he said. "We're saying that we want to improve them. And to improve them for African communities, we have to rejig the approaches we're using," says Adetunji.
Finding hidden high-risk groups
Beyond diabetes, the team believes that the new methodology could have wide-ranging applications across health research. The analytical approach could help identify hidden high-risk groups for other diseases, for example, by applying the same framework to different datasets.
"We're thinking that if people get to understand this method, how it works, it can be applied across the board," said Adetunji. "We just have to replicate what we've done with understanding or discovering these new subgroups of people that are at high risk for type 2 diabetes and plug it into another dataset with other health conditions."
The work also highlights the growing importance of African-generated data and locally driven data science expertise. Historically, researchers on the continent have had limited access to large-scale health datasets, often relying on evidence generated elsewhere.
"We've always relied on the Global North to give us solutions," Adetunji noted. "But now that we have the data, we can address our questions and develop appropriate analytical tools. The tools are where data science comes in."
Turning African data into African health solutions
He believes building data science capacity will be critical to unlocking insights from Africa's expanding health datasets.
"Building capacity to understand data science and increasing the number of data scientists could be the most optimal way to get more insights from the data that we have now," he said.
Looking ahead, Adetunji plans to continue exploring how data science and machine learning could uncover hidden disease risks and improve population health.
"I'm interested in trying to understand how we find more people who are at high risk for certain diseases and have been hidden for a long time," he said.
“This study demonstrates the potential of innovative analytical approaches to reshape health screening strategies and ensure that African populations benefit from contextual data-driven science,” says Professor Michele Ramsay, director of the Sydney Brenner Institute for Molecular Bioscience at Wits.