Health care has never had more data. The industry generates approximately 30 percent of the world’s data volume, and that share continues to grow faster than nearly every other industry. Electronic health records, claims systems, pharmacy records, lab results, care management notes, social determinants of health data, wearable devices, and patient-generated information are producing an unprecedented volume of information about the patients we serve.

At the same time, health care has never been more excited about artificial intelligence. Across the industry, organizations are launching pilots, building governance frameworks, evaluating vendors, and exploring how AI can improve everything from clinical decision support and care management to operational efficiency and financial performance. Yet despite the enthusiasm, many organizations are discovering an uncomfortable truth: AI is only as good as the data that powers it.

The industry has spent the last two years talking about models, algorithms, copilots, agents, and large language models. But the biggest obstacle standing between health care and meaningful AI transformation isn’t the technology itself.

It’s data fragmentation.

The problem isn’t a lack of data

Health care doesn’t suffer from a shortage of information. If anything, the challenge is the opposite. The average patient interacts with multiple providers, specialists, health systems, pharmacies, community organizations, and payers throughout their health care journey. Each interaction creates new data, often stored in separate systems with different formats, standards, and levels of accessibility. That fragmentation is structural: Health care organizations operate an average of sixteen or more distinct EHR platforms across their affiliated providers, according to HIMSS Analytics. While interoperability has improved significantly in recent years, nationwide exchange remains uneven, and many organizations still face challenges accessing complete patient information across disparate systems and care settings.

As a result, organizations frequently find themselves making decisions with only part of the picture. A physician may see clinical information from a recent visit but lack visibility into care received outside the network. A care manager may understand utilization patterns but not have access to important social factors influencing a patient’s health. A health system executive may review performance metrics without fully understanding the patient behaviors driving them.

When humans struggle to connect these dots, AI struggles too. The common assumption is that AI can somehow overcome fragmented information. In reality, AI often amplifies the strengths and weaknesses of the underlying data. If the data is complete, connected, and trustworthy, AI can generate powerful insights. If the data is fragmented, incomplete, or inconsistent, AI simply scales those limitations. The best model in the world will still fail without good data.

Context matters more than intelligence

One of the most persistent misconceptions surrounding AI is that increasingly sophisticated models will solve health care’s challenges. But health care is fundamentally a context problem.

A recommendation generated from claims data alone may look very different from one generated using claims, clinical records, laboratory results, medication history, and social determinants of health information together. The difference isn’t necessarily the intelligence of the model. It’s the completeness of the picture.

Health care leaders often focus discussions on explainability, governance, and model performance, and those conversations are important. But before asking whether an AI model is effective, organizations should ask a more fundamental question: Does the model have access to the full context required to understand the patient, provider, or population it is evaluating? Without context, even the most advanced AI system is making decisions with blind spots.

Why fragmentation creates risk

The consequences extend beyond missed opportunities. Fragmented data can create conflicting recommendations, inconsistent patient experiences, operational inefficiencies, and growing distrust in AI-generated insights. When clinicians receive recommendations that don’t align with what they know about a patient, confidence in the technology erodes. When care managers chase interventions based on incomplete information, resources are wasted. When executives receive different answers depending on which system they query, decision-making slows.

Trust is one of the most important currencies in health care. And trust in AI begins with trust in the underlying data. This becomes particularly important as organizations look to deploy AI in increasingly sensitive areas such as care management, clinical workflows, quality improvement, risk adjustment, and patient engagement.

The stakes are simply too high for incomplete information.

The longitudinal record becomes essential

To move beyond experimentation, health care organizations must shift their focus from AI tools to AI readiness. That starts with creating a comprehensive, longitudinal understanding of patients and populations.

Rather than viewing data as a collection of disconnected transactions, organizations need to create a unified view that brings together clinical, financial, operational, and social information into a single source of truth. When that foundation exists, AI can deliver far greater value. Clinicians gain access to relevant patient insights before an encounter occurs. Care managers can identify emerging risks before conditions escalate. Analysts spend less time gathering data and more time generating action. Executives can evaluate performance with greater confidence and precision.

The value isn’t created by the algorithm alone. It’s created by combining intelligence with context.

What we’re seeing across our customer base is clear: Organizations that have done the work to aggregate and connect their data are the ones best positioned to put AI to use. The gap between them and organizations still working from fragmented systems isn’t a gap in ambition. It’s a gap in readiness.

The future belongs to organizations that solve the data problem

Health care’s AI conversation is beginning to mature. The industry is moving beyond debating whether AI matters and toward a more important question: how to create meaningful value from it.

The organizations that succeed will not necessarily be the ones with the largest AI budgets or the most ambitious pilot programs. They will be the ones that recognize a simple reality: AI is not a substitute for strong data foundations.

In many ways, health care’s AI future looks remarkably similar to its data future. Success depends on connecting fragmented information, establishing trust, creating context, and ensuring insights reach the people who need them at the moment decisions are being made.

The excitement surrounding AI is well deserved. Its potential to improve health care is enormous. But realizing that potential depends on addressing a challenge the industry has struggled with for decades: data fragmentation.

Because in health care, intelligence without context isn’t intelligence at all.

Michael Meucci is a health care executive.