Industry Voice: Interoperability is the foundation for scalable, responsible AI in healthcare

For the past several years, healthcare has been flooded with AI pilots that promise transformation. Many of them work, at least in controlled environments. But across health systems, a different pattern is emerging. The challenge is no longer identifying use cases or selecting the right models. It is what happens after that decision is made.

As organizations attempt to scale AI beyond pilots, they are running into a more fundamental constraint: data.

AI is only as effective as the information it is built on. In healthcare, that information is often fragmented across clinical systems, devices, administrative platforms, and patient-facing applications. Even where digital infrastructure exists, data is frequently incomplete, inconsistently structured, or difficult to exchange across systems. This creates friction at every stage of care delivery, and limits the real-world impact of even the most advanced technologies.

This is not a new problem. For decades, the industry has worked to enable interoperability through shared data standards. Today, those standards, particularly HL7 FHIR, are widely used to support modern data exchange across healthcare systems worldwide.

But as AI adoption accelerates, the stakes are higher.

The conversation is shifting from whether data can be exchanged to whether it can be trusted, understood, and used consistently across workflows. AI introduces new demands on data quality, structure, and context. It requires not just connectivity, but alignment, so that data captured in one system can be meaningfully interpreted and applied in another.

This is where standards play a critical role.

By providing a common framework for how data is structured and exchanged, interoperability standards enable consistency across systems and stakeholders. They help ensure that information, whether clinical, operational, or patient-generated, can be used reliably to support decision-making, automation, and care delivery.

Importantly, this extends beyond traditional clinical systems.

Healthcare is increasingly becoming a distributed model, where care spans hospitals, outpatient settings, the home, and community-based environments. Data is now generated not only within electronic health records, but also through medical devices, remote monitoring tools, and consumer technologies. Integrating these data sources into a cohesive, interoperable ecosystem is essential to delivering more proactive, personalized, and efficient care.

AI depends on this broader, connected data environment.

Without it, organizations risk creating isolated solutions that cannot scale, or worse, systems that amplify existing inefficiencies due to inconsistent or incomplete data. With it, there is an opportunity to move toward more coordinated care models, improved patient engagement, and better use of limited healthcare resources.

At the same time, scaling AI in healthcare requires more than data exchange alone. It also depends on governance, ensuring that AI systems are transparent, explainable, and aligned with clinical and ethical standards. It requires lifecycle management, so that models can be monitored, updated, and improved over time. And it requires integration into real-world workflows, where clinicians and care teams can use these tools effectively without added burden.

These are not purely technical challenges. They are system-level challenges that require collaboration across the healthcare ecosystem.

That is why broad engagement is so important, from providers and developers to policymakers, researchers, and patients themselves. The future of healthcare will be shaped not by any single technology, but by how well we align around shared foundations.

Interoperability is one of those foundations.

It is what allows data to move securely and efficiently. It is what enables innovation to scale beyond individual systems. And increasingly, it is what determines whether AI can deliver on its promise in real-world healthcare settings.

The next phase of healthcare innovation will not be defined by new models alone, but by the infrastructure that supports them. Building that infrastructure, through open, standards-based approaches, will be essential to ensuring that AI is not only powerful, but practical, responsible, and ultimately beneficial to patients.