Programme will target African health sector workers looking to improve efficiency and outcomes
A South African NGO plans to set up a hub in Durban where people working in public health throughout Africa can come to learn how to use AI chatbots in their work.
THNK, a non-profit company founded in 2013 to help people affected by TB and HIV, says it hopes to have the hub up and running by mid-2024.
The training will build on methods to use ChatGPT to improve the efficiency and outcome of health programmes developed by THINK under a Gates Foundation funded seed project.
Last weekend, Kristina Wallengren, founder of THINK and its global CEO, presented the results of a pilot of the methods at a Grand Challenges conference in Senegal.
The pilot compared THINK’s AI tools with traditional ways of analysing South African healhcare data. AI was both faster and more accurate than the traditional ways, and requires less training to operate.
The pilot was carried out in partnership with South Africa’s health department.
Bright idea
Wallengren says she came up with the idea of using ChatGPT to analyse public health data when the Gates Foundation issued a call for proposals for ways chatbot technology can be applied for development in low- and middle-income countries.
Wallengren, who earlier in her career worked as an internationally mobile data-crunching healthcare consultant, tested the AI tool on the types of data and questions that she would have studied in that role.
She was amazed at how well ChatGPT could carry out the tasks she used to do, and the implications of this. The types of health consultancy she performed as a young epidemiologist cost low- and middle-income countries a lot of money, she said. That also meant it was rarely performed—perhaps only once every few years.
That was a major drawback, she said. “Everyone needs this information all the time. Not once every three years.”
Inspiring others
The hub will offer training to African health professionals in how to use AI to crunch their own data to improve efficiencies and detect gaps in healthcare provision.
The course will be based at THINK’s headquarters in Hillcrest in Durban. The technical requirements are simple, says Wallengren. “All we need is 30 computers and the ability to do hybrid training remotely.”
The training will target people who work in public health, such as district health administrators and programme managers. THINK will fundraise to be able to offer the training for free.
Much has been made of AI taking work from humans. But this initiative will not take work away from African data analysts, says Wallengren. Instead, it will equip public health systems in Africa with the analytical skills they used to have to bring in from outside.
She and her colleagues have also worked hard to ensure the methods THINK has developed are ethically sound and don’t jeopardise patient privacy.
The power of AI in public health is still unexplored, says Wallengren, and she would welcome other actors to join THINK in this space. “I’m hoping to inspire others. I believe in collaboration, not competition.”
