Fragmented coordination in AI implementation in health Implementing AI on a large scale requires a coordinated approach that engages both national governments and local champions. AI offers a unique opportunity to redesign health systems and workflows by integrating lessons from past initiatives while advancing new innovations. To do so, governments must exhibit expertise and leadership to create an appealing vision for AI in health and articulate a transformation journey to deliver this vision. Traditional change management often emphasizes operational or decision-making levels. In contrast, AI-driven change should be shaped by demand, incorporating feedback from end users, such as patients and communities, alongside input from payers and policy-makers. This dual-track approach ensures solutions are relevant, sustainable and scalable, thereby preventing fragmentation. By aligning these diverse perspectives, AI can be more effectively integrated into health systems, addressing real-world needs and ensuring broader adoption and success. Additionally, a public communications campaign is needed to ensure that AI-driven changes are seen to be beneficial for patient care and to encourage behavioural shifts, particularly when front- end pathways are affected (one example being digital front doors, which allow patients to access healthcare without the need to see a health professional). Change management processes will be more effective if designed and driven by local champions. Identifying and supporting these champions is crucial for successful implementation. In some countries, experts recognize that digital health operates on a push model due to low structural demand; in these settings, local champions are essential to ensure adoption and diffusion. For example, India’s AI for TB initiative has demonstrated significant success. By using AI- powered mobile apps, local health workers in rural areas have increased early detection rates of tuberculosis by 16%,15 showcasing how local champions can effectively use AI to improve health outcomes in underserved regions. However, they need to operate in a context where their push is paired with a demand or minimal pull. This demand is usually the responsibility of local policy-makers or decision-makers, meaning that the coordinated national governments and local champions change process is a requirement for scaling up AI in health. Another example is the National COVID-19 Chest Imaging Database (NCCID) in the United Kingdom, established by the country’s National Health Service (NHS) during the pandemic. This initiative collected an extensive repository of chest-imaging data (more than 40,000 X-rays, magnetic resonance images and computed tomography images) from all over the United Kingdom to support the development of AI tools for better diagnosis of COVID-19. The decentralized nature of the NCCID promoted collaboration among various NHS trusts, universities and private companies. More than 20 NHS trusts contributed imaging data, and several universities (such as the University of Cambridge) and research institutions used this data to develop AI tools. These tools were then made available at the local level to speed up the process of identifying patients at risk of severe complications, enabling quicker interventions and improving intensive care unit resource allocation.16 Short-term pressures and long-term sustainability need to be balanced Overcoming these challenges and securing AI’s place in healthcare requires a strategic approach that balances short-term political pressures with long-term financial sustainability and innovation. Even with these solutions in place, success hinges on leadership that actively engages with the technical decisions critical to AI integration. Too often, leaders defer these crucial decisions to technical experts, creating a risk of fragmented and inefficient systems. To ensure AI adoption is both strategic and sustainable, it is essential for leadership to take an active role in shaping the technical landscape and guiding the integration process effectively. i The Future of AI-Enabled Health: Leading the Way 17

The Future of AI Enabled Health 2025