Using AI to tailor drugs for Africa

Therapeutic drugs have long been ill suited to African patients’ needs. Professor Kelly Chibale, the director of the University of Cape Town’s (UCT) Holistic Drug Discovery and Development Centre (H3D), and his research colleagues posit that scientists can use artificial intelligence (AI) to change this.

Published in the 11 April 2024 edition of Nature, “AI can help to tailor drugs for Africa – but Africans should lead the way” outlines how machine learning (ML) can improve our understanding of how therapeutics might affect patients of African descent.

The research alluded to in the article has been undertaken as part of Project Africa GRADIENT (Genomic Research Approach for Diversity and Optimising Therapeutics). This initiative is aimed at understanding the variable genetics on the African continent and the impact they have on variable response to medicines in African populations.

Changing the DNA of therapeutics

Professor Chibale and his co-authors’ inquiries were driven by the understanding that drug research and trials often fall short both at the clinical and preclinical phases, resulting in the development of medicines that aren’t necessarily fit for treating African patients.

“Today, Africa makes up 15% of the world’s population, but carries 20% of the global disease burden. Although we have this scenario, medicines haven’t historically been optimised for the African patient population.

“This is partly because there has been a very low volume of clinical trials on the continent. On average, about 3 to 4% of global clinical trials happen in Africa. This means, by implication, that the therapeutics coming out of these trials are optimised largely for people from the developed world or at least outside of Africa,” he explained.

“Today, Africa makes up 15% of the world’s population, but carries 20% of the global disease burden.”

“Even before those investigations, however, preclinical trials that study how we metabolise specific drugs, use tools – for example, liver cells containing drug metabolising enzymes – from Caucasian donors.

“No livers from African donors are used in the conventional discovery process, the absence of which means that the predicted dosages in clinical trials do not account for the massive genetic variation that we know exists in the African population.

“Genetic differences in the expression and activity of drug metabolising enzymes and transporters ultimately result in variable responses to therapeutic drugs.”

In this vein, Project Africa GRADIENT is critically important, not only for driving the development of therapeutics that are effective in treating diseases, but also optimising the dosage that patients receive.

“Project Africa GRADIENT is an initiative to begin to understand what the impact is of this huge genetic variability that exists in Africa in terms of our response to medicines,” Chibale added.

Fast-tracking medical research with AI

Traditionally, collecting the data required to advance a study like this on the African continent has been challenging.

Now, AI presents an opportunity for researchers to use the limited data they have at their disposal to build models that can effectively predict the potential outcomes of therapeutic treatments.

Within the context of the Project Africa GRADIENT initiative, H3D in collaboration with Ersilia Open Source Initiative (EOSI) have been using AI to identify genetic variants that are prevalent in Africa and likely to affect the metabolism of malaria and tuberculosis (TB) drugs.

These genetic variants of interest are being incorporated in existing mathematical models to come up with proposed tailored dosages, which need to be evaluated in ethnobridging human clinical trials.

“AI has a lot of potential to accelerate medical research in Africa.”

“AI has a lot of potential to accelerate medical research in Africa, but there are several barriers for the realisation of its full potential, including access to affordable power or electricity, connectivity or digital infrastructure, and data,” Chibale said.

“When it comes to data, you need rich, highly intensive, granular data to do these types of analyses. Unfortunately, this has been lacking on the African continent. There hasn’t been effective collection of data and there is little access to the data – especially in a format that allows us to use AI and ML.

“However, because of the potential impact of this research, we can’t wait until we have all the data that we need. So, the approach that we’ve taken is to recognise that there is some data that we can work with and then to use transfer learning from other areas to leapfrog over this obstacle.”

According to the director of H3D, this is something that African researchers must take advantage of. In this instance, it’s not only the synthesis of more appropriate therapeutic treatments, but also ensuring that Africa can capitalise on these research methods that should be a motivating factor.

“We’re beginning to see applications of AI in almost every sector of society, but we are already seeing that we are lagging behind. Unfortunately, if we are not careful, this chasm that has begun to develop between Africa and the rest of the world will continue to grow and we will be left behind,” he explained.

The future depends on African innovation

With myriad challenges to overcome – think access to electricity and internet services along with inadequate data collection – Chibale noted that it’s essential that African scholars lead the way in this research.

“There is a very strong link and correlation between the genetics of the population, the social and physical environment in which those patients live, and treatment of disease. Therefore, it’s a no-brainer, that doing the discovery and the development in close proximity to where the patients are is a better way to address unmet medical needs of those people.

“Without understanding the social and physical environment in which these patients are and the researchers understanding the communities that they’re going to be serving, innovation will be pointless,” he said.

“The disease burden falls largely on the shoulders of Africans. It’s mostly us, so this will likely not be a priority for other countries or regions; we must take matters into our own hands to secure a better future on our continent.”

“There is no question that we as Africans have been beneficiaries of the health innovation that has taken place in the developed world.”

What’s more, Chibale highlighted that innovative contributions to research can help Africa to take its place on the global stage.

“There is no question that we as Africans have been beneficiaries of the health innovation that has taken place in the developed world in terms of the discovery and development of innovative medicines.

“We must now contribute something in this area as we have been doing in other areas as exemplified during the COVID-19 pandemic. We can’t just be beneficiaries of other people’s health innovation, we must also innovate and bring something to the table.”

Project Africa GRADIENT is an initiative of innovative pharmaceutical companies GlaxoSmithKline (GSK) and Novartis in partnership with the South African Medical Research Council (SAMRC). The article “AI can help to tailor drugs for Africa – but Africans should lead the way” was authored by Gemma Turon (EOSI), Mathew Njoroge (H3D), Mwila Mulubwa (H3D), Miquel Duran-Frigola (EOSI) and Kelly Chibale (H3D).