The University of Cape Town has won $ 1.6 million, about R26.4 million, from Google.org, Google's philanthropic arm, to build an AI platform that maps how genetic differences in African populations affect the way patients process medicines. UCT is one of 17 organisations worldwide selected for the Google.org Impact Challenge: AI for Science, from more than 3,000 applicants in 30 countries. The 17 will share $ 36 million, roughly R608 million.

The platform is called AfriPGx-AI, and it tackles a gap in pharmacogenomics, the study of how a person's genes affect their response to medicines. The same drug at the same dose can work well for one patient and cause side effects or fail in another, partly because of genetic differences in how the body breaks it down.

Most prescribing guidelines that account for this were built almost entirely on European and East Asian data, so doctors in Africa often prescribe without evidence from their own populations. "Medicines do not work the same way in everyone, yet the evidence that guides prescribing in Africa has largely been generated elsewhere," said Professor Kelly Chibale, founder and director of H3D, UCT's drug discovery centre. "Africa is home to the greatest human genetic diversity on Earth, and that diversity must be at the centre of the science that shapes how medicines are used."

The team will use AI to catalogue genetic variation across more than 50 genes involved in processing medicines in African populations and predict which variants are most likely to change how a drug is metabolised. The strongest predictions will then be tested in the laboratory. It starts with medicines for tuberculosis, HIV and malaria, and the decision-support tools are being designed to work in low-bandwidth settings.

The software, AI models and datasets will be released openly. Chibale leads the work alongside Dr Musalula Sinkala, Dr Mwila Mulubwa, Dr Mathew Njoroge and Professor Jonathan Shock, interim director of the UCT AI Initiative, and the award will also train African postdoctoral and postgraduate researchers.

IT-Online reported the award and the project's plan to test AI predictions against laboratory results. UCT News carried Chibale's full explanation of why pharmacogenomic research in African populations is so limited. Google's announcement lists all 17 recipients, with UCT among the health and life sciences winners. Google.org's Leslie Yeh said UCT was combining AI and deep scientific expertise in ambitious, creative ways.

The core problem here is one we keep running into: AI and medical evidence built on data from somewhere else work worse here. Injini made the same argument about education when it said Africa is underrepresented in the data that trains AI tools. UCT's project goes at the gap directly, building the data from African populations, in Africa, led by African scientists. Two details make it more useful than a typical research grant.

It starts with TB, HIV and malaria, the diseases where getting a dose wrong affects the most people on the continent. And the tools, models and data will be released openly, which means local healthtech companies and other researchers can build on them rather than start from scratch.

That fits the opening created by new rules letting more health professionals work independently: clinical decision tools designed for South African patients have a growing market. R26 million will not change prescribing on its own, but it pays for the evidence base that everything else depends on.

You might also like our piece on the first AI-assisted brain tumour operation, the Wits topological light breakthrough, and Navi Pillay winning the Nobel Peace Prize.

Get more SA tech and business news and subscribe to The Open Letter.