• News
  • September 18, 2018
  • 6 minutes
  • 0

The global south is getting its own AI healthcare revolution

AI in healthcare is predicted to save the US $150bn annually by 2026

Artificial intelligence in healthcare is no longer the preserve of rich countries — it could be on the verge of transforming medical care across the global south.

A new report, published in the British Medical Journal in late August, sets out how AI is already changing healthcare practices from rural Brazil to sub-Saharan Africa.

“It’s unprecedented,” said Brian Wahl, a researcher at Johns Hopkins Bloomberg School of Public Health, and author of the paper. “We’re entering a time where AI could have impacts in developing countries similar to the impacts it’s having already in developed countries.”

• For more like this, see our public health newsfeed.

Research published last year found that AI could save $150billion in annual healthcare costs in the US alone by 2026.

The AI revolution is still yet to fully develop, Wahl cautioned, but the penetration of smartphones into low-income settings, developments in cloud computing and the digitisation of health records as standard practice have built the framework on which artificial intelligence relies.

The types of AI being developed in the global south differ from those in the north, however.

“Using AI in a networked hospital system in the US is different from providing care in the community in a resource-poor setting. In rural parts of poor countries, there’s a lot of potential for AI technology to support the work of community health workers providing care outside of health facilities and in the community,” Wahl said.

One promising avenue is in the use of AI to more efficiently schedule appointments for community health workers who provide support to communities without healthcare centres or hospitals.

In Bangladesh, community health workers have helped to boost neo-natal survival rates by over 30%, despite a high caseload and long distances to travel that meant workers were present at only 5% of births.

In sub-Saharan Africa, extensive digitisation and data collection helps community health workers organise their caseload and schedule appointments more efficiently to avoid gaps in service. And a mobile application in Tanzania helps community health workers deliver more targeted information to pregnant women and young caregivers.

Meanwhile, researchers in Brazil and the US developed a machine learning application to predict the need to resuscitate newborns suffering from birth asphyxia, a condition which killed 800,000 newborns in developing countries every year in the 1990s.

Rigorous evaluation has deemed the application highly accurate, and a promising diagnostic tool in resource-poor settings.

But the promise of AI in rural and low-income settings comes with costs.

“A challenge with AI applications is that they’re a “black box”: data goes in, it gets crunched in a  certain way that’s rather opaque, and there’s always the risk of biases — intentionally or unintentionally — creeping their way into the technology,” said Wahl.

Those concerns are common in the global north, but resource-poor settings can present their own challenges, which require tech developed in rich countries to be comprehensively adapted to its new setting.

Applications of artificial intelligence in low-income countries must be developed in line with local cultures and laws, Wahl said. A diagnostic tool must function under the same regulations as healthcare professionals, for example, which will differ from country to country. And tools that recommend treatments unavailable due to resource constraints are unlikely to be of use in many developing countries.

Then there’s the question of ownership and privacy. The lure of outsourcing AI development to private companies is greater for cash-strapped governments for whom erecting a digital infrastructure will always lose out to the daily challenge of basic service provision. That could jeopardise the ability of poor nations to hold big tech to account when things go wrong.

Activist charity Global Justice now has already warned of an “e-pocalypse” as tech giants move to monopolise new African markets.

But regulatory frameworks are in development: the Nairobi Data Sharing Principles, written in 2014, are a landmark attempt to develop minimum data protection standards in the global south. Uptake remains sluggish, however.

For better or worse, AI is coming to the developing world, said Wahl.

“AI could help overcome some basic challenges in health systems in developing countries, and we need to be talking about it.” — Edward Siddons

(Picture credit: Flickr/BTO)


Leave a Reply

to leave a comment.
Master the skills you need for the public service.

Discover inspiring resources, tools and policies.