Experts Highlight Importance of Custom-built AI System in Africa’s Healthcare Delivery System

Emma Okonji

Medical and communications experts have stressed the importance of widespread adoption of Artificial Intelligence (AI) technology in healthcare delivery across Africa.

They however highlighted the need for a custom-built AI system that would promote trust and credibility among medical practitioners in healthcare delivery across Africa.

They spoke at a webinar, organised by The Newmark Group, a leading Pan-African Integrated Marketing Communications firm, themed: ‘AI in Healthcare: Opportunities And Challenges’, which was moderated by the Country Lead, Newmark Ghana, Mr. Wilfred Kedapey.    

Co-Founder, Rx Health Info Systems, Mr. Daniel Marfo, urged African medical practitioners to consider working with AI tools that are customised or built along some treatment plans and guidelines in the country.  

According to him, at the macro level, where decisions are being made by policymakers, and decision makers, we are seeing AI systems sitting at the heart of all the data culminating in the health systems of countries coming through, and they are using systems ranging from cloud AI to their own custom-built AI systems to be able to review and assess the macro picture.

Speaking about the importance of AI in healthcare delivery systems across Africa, Marfo said: “The way AI is able to analyze x-ray, lab results, and images, is transformational. Things that you may have missed, if you were using your eye to study an x-ray or an MRI or a CT scan, the AI today is able to help identify all those things, and it’s very instrumental because today there are a very limited number of radiologists and specialists all over Africa who are available to review medical results.”

The other trend we have seen is that a lot of hospital management systems are now really coming fully embedded with AI systems to assist the healthcare in a manner that will make them work faster, better and become more professional. Today, AI systems are in the position to process hundreds of thousands of claims per day in the medical field, Marfo further said.

“Once you work with AI systems, which are built on the knowledge base of Africa, of specific treatment protocols and treatment guidelines in those jurisdictions, then when you are already infusing it, first you have great confidence from the ministries of health, from the supervising agencies, and from the medical teams within the hospital, because they know that this is actually giving them, and helping them with everything, but within the context of how they work,” Marfo added.

Addressing the operational barriers in implementing AI across African healthcare settings, the Founder, Aduro Analytics, Dr. Afriyie Karikari Bempah, said some of the operational barriers were mainly about having the AI being trained on local data and the protection of patient data.
“You’re not just looking at consequences that will affect reputation, but you’re also looking at lives. And in applying AI in healthcare, we are looking at very sensitive information that has to be protected. So the best operational barrier for me is making sure that patient rights and patient information are always protected.
As long as you can overcome that hurdle, it makes subsequent conversations easier,” Bempah said.

CEO, The Newmark Group, Mr. Gilbert Manirakiza, said to ensure that the data of patients and other stakeholders in healthcare is safe, is something very critical.
“I believe clinical institutions or even governments or ministries should build custom-made AI-powered platforms that can then be trained on the locally relevant data, because bias is a real and structurally significant issue, not just in healthcare, but across many other sectors. And that is essentially stemming from the fact that a lot of the large language models that we are using now are trained using populations, languages, cultures, education levels, and geographies that are not African,” Manirakiza said.

AI tools are trained predominantly on Western and often English language datasets, and these do not automatically understand Africa’s own local nuances and architectures. Therefore, we have to think about this very critically because when AI gets healthcare communication wrong in this context, the consequences are not just reputational, they are also human, which brings us to a very important question of trust, Manirakiza further said.