EVAH: a new initiative to generate evidence for AI in health
To make AI tools for health safe, effective and equitable, we need strong evidence from the settings where they’ll be used. We're co-launching a new initiative to support locally led evaluations of these technologies in low- and middle-income countries.
Credit:
Licence: All Rights ReservedNitat Termmee via Getty Images
5-minute read
5-minute read
Artificial intelligence (AI) has the potential to strengthen health systems. From diagnostics to optimising clinical workflows, AI tools could offer affordable solutions to address healthcare inequities, particularly in low- and middle-income countries.
But there is limited evidence about how well AI tools work in practice, making it difficult to know which innovations are ready to adopt and scale up.
We’re launching the Evidence for AI in Health (EVAH) initiative, in partnership with the Gates Foundation and the Novo Nordisk Foundation, to help change that.
Through a joint investment of US$60 million, the initiative will support the evaluation of AI tools for health in low- and middle-income countries over the next three years.
Why does Evidence for AI in Health (EVAH) initiative matter?
We need strong, reliable evidence before introducing AI into healthcare. This ensures these tools are safe, effective and able to improve patient outcomes across diverse populations and clinical settings.
Despite the vast expertise and innovation in low- and middle-income countries, there is a persistent gap in evidence generation for AI tools developed within these settings. Local and regionally designed solutions often lack the financial and structural support needed to evaluate tools.
EVAH aims to change this by strengthening local capacity to evaluate tools and investing in the evidence that guides which technologies will be adopted and scaled. Local validation is crucial to mitigating risks that can worsen, rather than reduce, health disparities.
“Only by working in partnership, and investing in rigorous evidence generation and learning, will we be able to support decision-makers and services to meet the needs of the communities they serve,” says Charlotte Watts, Executive Director of Solutions at Wellcome.
What AI tools for health will EVAH evaluate?
EVAH will work with country partners to select tools for evaluation that align with local health priorities and can be integrated into primary and community healthcare settings. The tools will span a wide range of AI technologies, such as:
- prediction models – for example, tools that can predict disease risk or prioritise patients for follow-up based on clinical history
- computer vision – for example, tools that can analyse visual information like X-rays or ultrasound scans
- large language models – for example, tools to support health workers with decision-making or clinical documentation
- multimodal AI – for example, tools that combine data types such as visual, audio and electronic health records to build a better understanding of patient risk
Technologies that are designed for resource-limited settings and trained on data that accurately reflects the populations they are intended to operate in will be given priority.
Tools will be evaluated through a range of methods, including:
- implementation research – studies that assess how an AI tool works in real-world situations
- randomised controlled trials – comparative studies that measure the effectiveness of a tool by randomly assigning participants to use it or not
- economic evaluations – analyses of the costs and potential health benefits to help decision-makers determine value for money
- acceptability studies – research on how patients, clinicians or communities who will use these tools perceive them
The first requests for proposals will enable locally-led evaluations of AI-enabled decision support tools. The tools should be ready for use and designed to assist frontline health workers in Sub-Saharan Africa and South and South-East Asia with clinical tasks, such as triage, diagnosis or referral.
This is run in partnership with the Abdul Latif Jameel Poverty Action Lab and the African Population Health Research Centre, two organisations with strong experience in carrying out high-quality research and analysis that informs policy in low- and middle-income countries. They’ll be involved at every stage, including scoping the tools, developing evaluation frameworks, summarising evidence and disseminating findings to regional and international policy experts.
Driving better health outcomes for all
At Wellcome, we believe how research happens matters just as much as what it seeks to answer. Engaging with people who are most affected by the issues we want to help solve will unlock the greatest long-term impact.
This is especially important for AI tools and systems, which reflect the data they are trained on.
“Ensuring new AI tools are backed with real-world evidence can significantly reduce the time it takes to turn promising ideas into scalable innovations,” says Trevor Mundel, President of Global Health at the Gates Foundation.
We recognise that philanthropy has a big role to play in how AI is designed and deployed to advance health. But we can’t do it alone.
Real progress needs governments, researchers, industry and communities to work together to ensure AI delivers meaningful and equitable benefits for all.
That’s why all findings from EVAH will be free to access online, making the evidence available to anyone working to strengthen health systems or develop better AI tools. This will be underpinned by standards that prioritise data privacy, safety and ethics.
Additionally, through our partnerships with the Abdul Latif Jameel Poverty Action Lab and the African Population Health Research Centre, we’re collecting and sharing evidence in ways that support local leadership and build trust and engagement with these tools.
Publishing the findings through open access channels could accelerate the impact of the AI tools, explains Lene Oddershede, Chief Scientific Officer, Planetary Science and Technology at the Novo Nordisk Foundation. These evaluations, she says, will “provide decision-makers with crucial data on efficacy, economic value and acceptability of these technologies in the contexts where they’re most needed”.
Supporting health research in Africa and Asia
The Evidence for AI in Health (EVAH) initiative builds on our long-term support for health research in Africa and Asia. We invest in people and partnerships across these continents to support local research for regional and global impact.
In addition, we fund research in Africa and Asia through our four strategic funding programmes. These are Climate and Health, Discovery Research, Infectious Disease and Mental Health.
Our funding supports early-career, mid-career and established researchers across a broad scope of research, including physical and social sciences, clinical research and humanities.
Tags
