Last month, Economist Impact’s Policy & Insights Health Team took the global digital pulse of artificial intelligence (AI) at the heart of the future of vaccination—the 2024 World Vaccine Congress (WVC). Our team not only made an impact left an indelible mark by presenting key findings from The Vaccine Ecosystem and moderating key panels, but also took stock of the latest trends and innovations at the forefront of global health. Not surprisingly, AI and data emerged clearly as important topics when considering their role in global health responses. From the extensive discussions and expert panels, three key takeaways highlighted the cautiously optimistic expectations of AI as a force changing the global health landscape.
1. AI's current limitations in global health prioritisation
AI’s $16.3bn market in healthcare is expected to grow by 40.2% by 2029, indicating strong competition between traditional players and non-health technology companies to offer highly integrated and personalised solutions, as suggested by our latest research. Recognising this trend, expert panellists from the World Health Organisation, the Coalition for Epidemic Preparedness Innovations and the National Institutes of Health discussed the role of AI in shaping global health priorities at the WVC, with a focus on aiding in drawing up global priority lists for threatening pathogens and tropical diseases.
The discussion pointed to significant challenges in leveraging AI to assist with global strategy and identifying priority pathogens, primarily due to the quality and quantity of the available data. Issues such as delayed reporting, inconsistent data cycles and the lack of digital infrastructure significantly undermine the potential of AI. These gaps hinder AI's ability to effectively prioritise health resources where they are most needed, such as in regions where Lassa fever and Malaria are endemic. While AI holds promise, it must still be appropriately equipped to handle the complex tasks of assisting experts with formulating global health prioritisation lists, requiring data and algorithm refinement and auditing. Our research indicates that validation is key, as lives will be at stake.
2. The imperative of collaboration and partnership
The data divide presents a considerable challenge for training AI algorithms, with data gaps found particularly across genders and underrepresented populations. One critical step to bridging this data divide is collaboration, as pointed out by Dr Mandy Cohen, director of the US Centres for Disease Control (CDC) and Prevention, in her keynote session. Dr Cohen emphasised the critical role of data partnerships in improving global health responses by breaking down silos and focusing on partnerships that are results-oriented to enhance data collection and sharing.
This is exactly the challenge that a study in the Journal of Infection and Public Health pointed out: low- to middle-income countries are struggling to break down data silos, often hampered by limited infrastructure, insufficient training and organisational limitations. In the US, the CDC has championed its role in digital transformation and modernisation of data collection, such as expanding electronic laboratory reporting, real-time data provision and multisector partnerships, which now enable quicker responses to public-health threats. Goal-oriented collaboration is essential for effectively building the comprehensive data infrastructure required to tackle future health challenges and improve the data flow, thereby overcoming fragmented systems, as seen in Europe during the covid-19 pandemic.
3. The paramount importance of transparency for trust
Transparency should be at the heart of collaborative data collection, processing, reporting and use of AI solutions—a concept emphasised by a keynote panel addressing gaps in adult immunisation. One example highlighted during the session was Vaccine Track, a platform available to key public-health stakeholders that uses claims data of routine adult vaccinations to bolster transparency and access to immunisation data. However, transparency is not just about data accessibility but also about understanding how AI models are designed to function and make decisions.
Systemic and subconscious bias in algorithmic design and training must be addressed, and can be uncovered through transparent collaboration among different stakeholders. Our research indicates that the inclusion of underrepresented groups and communities is essential during the technology-design phase to increase engagement and develop inclusive technologies for all. To use AI to assist with decision-making processes, it is critical to ensure transparency in AI processes to help mitigate biases and foster deeper trust in the technology. This trust is vital for its acceptance and integration into health strategies and interventions.
Moving Forward
We should be cautiously optimistic about integrating AI into decision-making processes, a concept widely discussed at the congress. While AI and data offer incredible potential to transform global health and vaccine development, significant challenges continue to overshadow optimism. However, there was a clear call for collaboration at the international level to take the necessary steps to address these concerns. Stakeholders need to strategically target hurdles associated with data integrity by fostering collaborative frameworks and maintaining transparency in AI algorithms. Addressing these challenges is pivotal for effectively deploying AI tools to improve global health outcomes.
