There is currently a significant amount of buzz in global health regarding artificial intelligence (AI) as it creates new opportunities to improve health and wellbeing on a global scale in ways
that humans alone cannot. A recent report sponsored by the United States Agency for International Development (USAID) and the Rockefeller Foundation in collaboration with the Bill and Melinda Gates Foundation, Artificial Intelligence in Global Health: Defining a Collective Path Forward, highlights 27 use cases, eight challenges, and six investment priorities aimed at capitalizing on newfound technological capabilities to improve health and wellbeing in low and middle income countries. The report contextualizes AI within the broader umbrella of digital health and identifies many of the same enablers that need to be addressed by countries for investments to succeed, namely:

1. Data availability and quality
2. Business model and sustainability
3. Privacy, ethics, and ownership
4. Regulations and policy
5. Integration into health system
6. Required evidence of positive impact
7. Gaps in AI building blocks
8. Gaps in required infrastructure

Each of these challenges has a corollary to the broader domain of digital health that if addressed would create a winwin for the effective use of digital technologies and the data they generate and lay a solid foundation for the effective use of artificial intelligence in global health. The State of Digital 2019 Report provides a global trends analysis and showcases the current state of digital health maturity in 22 countries participating in the Global Digital Health Index, providing a high-level barometer indication of adoption readiness that can be applied to artificial intelligence. The majority of participating countries have achieved only average maturity in their digital health systems – indicating room for growth with most falling below average in particular areas of relevance to AI. As a region, Africa is the furthest behind, but stands to gain the most from artificial intelligence if approached responsibly. Throughout the world, the main digital health enablers with low overall maturity are nationally scaled digital health services and applications, digital health architectures to facilitate interoperability, privacy policies, regulations supportive of connected health, and a digitally ready health workforce.

In addition, noteworthy risks of AI in global health are the same as those for AI in general, but unaddressed they have the potential to do more harm than good when amplified at a global level and deployed in resource — constrained environments that cannot afford high margins of error. Its application to low and middle-income countries, where technology literacy and programming is male dominated and health data is sparse, raises key questions related to which human brains machines are trained to mimic, evoking important considerations related to gender and racial bias in the underlying algorithms and data inputs that form the basis of automation and insights. To build a solid foundation for AI in low- and middle-income countries, the following actions will be needed: Raise awareness among policymakers and build capacity of the health and digital health workforce on how best to integrate AI in a way that is both responsive to the local health priorities and responsible in engaging people and protecting their rights. Promote the national scale-up of digital health services and applications to generate the volume of local data needed to power algorithms that produce locally relevant insights and action.

Review and update current privacy and personal data ownership policies to ensure that people are made aware of how their personal data are being collected, stored, and used, including secondary uses of data beyond the initial purpose of its collection. Adopt regulations that promote the use of connected health devices to facilitate capture of more biometric data from
individuals in low- and middle-income countries in support of movements towards AI to generate real-world evidence and build a local evidence base faster. Focus on the building block of machine learning with direct engagement of representative stakeholders that reflect the gender, racial, geographic, and linguistic profile of the population in data inputs and algorithms.

If approached proactively and responsibly locally and nationally, AI can be a powerful tool globally for predicting and controlling epidemics, augmenting and supporting the currently stretched health workforce,empowering individuals with the information they need to take a more proactive role and make better informed decisions about their own health, and facilitating resource optimization in health systems.

Patricia N. Mechael,
PhD MHS is Co-founder and Policy Lead at HealthEnabled

AIM Magazine Volume 2 Issue 3