Turning AI's Hype into a Realistic Hope for Global Healthcare

Amid all of the hype about how artificial intelligence may reshape healthcare around the world, Moka Lantum, M.D., is worried about a different question. For the Kenyan physician and entrepreneur, a more important issue is how should AI reshape healthcare delivery and access for under-resourced systems like Kenya’s.

Lantum is the founder and CEO of CheckUps Medical, a tech platform providing healthcare information and services to 300,000 people across Kenya and South Sudan. During a recent panel discussion at the Duke Global Health Institute, he cautioned that while AI holds great potential to expand access to health for people in his country, it could also make inequities in care worse. He noted, for example, that very few healthcare facilities in low-resource settings are even using digital medical records.

“The commitments we need to make in healthcare have not yet been made,” Lantum said during the event. Absent of those commitments, we will fall short of leveraging AI to the extent to which it needs to be leveraged. We have to be very deliberate about what we should do with AI, and not just what we can do with AI.”

The Oct. 1 event, titled “AI in Global Health: Hope, Hype and Realities,” featured four experts on AI and healthcare innovation, who considered the potential of emerging AI technologies from a global health perspective. All expressed enthusiasm for the power of AI-driven technologies to help healthcare workers make better-informed decisions and provide tools in places where access to healthcare professionals is limited. But they noted considerable challenges to ensuring that new technologies reach the communities that can most benefit from them.

“I actually share a lot of the hype in terms of the potential of AI for transforming healthcare, including in low-resource settings,” said Wendy Taylor, president and CEO of the William Davidson Institute, a nonprofit affiliated with the University of Michigan, and chair of DGHI’s Board of Advisors. She noted the potential of AI to improve prediction and response to infectious disease outbreaks and to deliver personalized medicine based on patients’ individual risk factors.

But she also noted that many promising innovations have failed to reach a scale of adoption to deliver systemwide gains. “One of the things I’ve learned from making investments in lots of innovations over the years is that the innovation part is actually the easy part, Scale is what’s hard,” she said.

Ruchika Singhal, president of Medtronic Labs, talked about how her company is using AI in digital platforms used by community healthcare workers in sub-Saharan Africa and South Asia. She said AI is aiding health workers in screening patients for conditions such as diabetes and cardiovascular disease by filling in gaps in patient data, helping them determine which patients should be referred for care.

The North Star for all of our work has been the value,” she said. “What is the problem we are trying to solve? Can we measurably move the needle on that problem and eventually demonstrate the economic benefits?”

Michael Pencina, Ph.D., Duke Health’s chief data scientist and director of Duke AI Health, shared the optimism about the use of AI tools to inform healthcare decisions. But he also noted that the accuracy of those decisions will depend on the validity of the data used to train AI decision models. He noted one example of how a biased measure of how often Black Americans go to the hospital led an AI tool to significantly understate their need for health services.

“It’s not the algorithm that was wrong. It was how humans built it and put it together,” he said. “It’s a reflection of what it was fed, and I think that’s important to keep in mind.”       

Watch the full event above or scroll down for highlights.

Speakers

Krishna Udayakumar, M.D., (moderator) is the founding director of the Duke Global Health Innovation Center and the executive director of the not-for-profit Innovations in Healthcare. He serves as an associate director for innovation at DGHI.

Moka Lantum, M.D., is the CEO and Co-Founder of CheckUps Medical, a platform providing tech-enabled urgent care and home delivery to more than 300,000 uninsured and vulnerable patients in Kenya and South Sudan.

Michael Pencina is a professor of biostatistics and bioinformatics at the Duke University School of Medicine. He co-leads the national Coalition for Health AI, whose mission is to increase trustworthiness of AI.

Ruchika Singhal is a seasoned healthcare industry leader with 20-plus years of expertise in health technologies and systems innovation, market development entrepreneurship and strategy across the globe.

Wendy Taylor is president and CEO of the William Davidson Institute, a non-profit affiliated with the University of Michigan. Taylor is also chair of the DGHI’s Board of Advisors. As Vice President for Technical Leadership and Innovation at Jhpiego, she led a team focused on driving accelerated impact across areas such as women’s health, infectious disease and climate-health.

 Highlights

The role of AI in developing sustainable solutions

“We are 10 years since the adoption of electronic medical records (EMR), and the EMR transition that was supposed to happen really hasn’t. Only 13 percent of healthcare delivery settings are digitized so that means [many] are excluded from this conversation. The challenge for AI, is enormous.”

Moka Lantum

“Is it going to lead us astray or do something malicious or problematic?  There’s an example of this which took an algorithm… to decide which patients need extra preventative health to keep them out of the hospital. In this case, Black Americans didn’t look like they needed much interaction with the healthcare system, so the algorithm said 17 to 18 percent need that help when it’s closer to 50 percent. It’s an enormous amount of bias. It’s not the algorithm that’s wrong, but how humans built it and put it together.”

Michael Pencina

What must be in place for AI to work

“The biggest thing is integration into health systems, and why innovation fail to scale over and over again in health at large. There’s this cultural piece of clinicians are trained a certain way and [they have] all the knowledge. I’ve heard clinicians say they don’t want to use it because it makes them feel less than. So how do we think about crossing that bridge so it’s a tool that’s an enabler and not a threat to their credibility?”

Wendy Taylor

“We use AI in operations to affect every way of patient communication, optimize decision making and booking for productivity. Implementing it at the limited scale we’re doing, if you do not have a workforce that is trained for data collection, you’ve lost the bet from ground zero.”

Moka Lantum

What kind of change can AI provide in the future

“Especially in improving access, we’re never going to have enough doctors, health systems and if these AI tools [can] improve outcome and value, that would be a success. If we can shift the needle on outcomes at a country level, that would be amazing.”

Ruchika Singhal

“There’s so much potential to take high quality interventions into lower-level facilities and increase access to care. What if you have a new technology [with] thermal imaging, it’s handheld and it’s AI-based where you can detect cancer early, and you don’t need [someone with a] Ph.D., M.D. and decades of training; the device tells you. That helps us address [workforce] gaps and shortages and can increase access to much better care.”

Wendy Taylor

 I don’t think a healthcare system where we’re waiting to see a PCP (primary care provider) for three months is okay. AI has no future in that reality. We have to get care back in the home, so a mother can set something on her counter and say, ‘Is this strep throat? Can I send my kid to school today or keep them home?’ "

Moka Lantum