AI can’t save lives if our data stay broken — A wake-up call for U.S. primary care physicians

© sdecoret - stock.adobe.com

Every day, U.S. primary care physicians and clinicians are expected to do more with less: more patients, more chronic conditions and more paperwork, and often with less time, support and clarity. Against this backdrop, it’s no wonder that artificial intelligence (AI) is being touted as the great clinical equalizer: a tool to reduce administrative burden, unlock better insights and deliver more personalized care.

But here’s the truth you may already know: AI can’t help if the data it relies on are fragmented, outdated or locked away.

Right now, this is the state of health data in many practices, clinics and hospitals across the country.

Let’s get clear about what’s holding us back, and what we can do about it.

The burden of broken data

Errol Rodericks
_{© Denodo}

Physicians are already drowning in data, but not the right kind.

Patient histories, lab results, prior authorizations, pharmacy records, specialist referrals, social determinants of health (SDOH) and claims data are often scattered across siloed systems. It’s not just hard to find what you need — it’s hard to trust that it’s complete, up to date or even accurate.

For example:

• You get a patient transferred from another clinic — but her records are stuck in a different system.
• You want to leverage an AI assistant to summarize a case, but it pulls from incomplete or outdated notes.
• You prescribe a medication, but you don’t see the recent test result that might have changed your decision.

In each of these cases, it’s not the AI that’s failing, it’s the data foundation.

The prior authorization wake-up call

Recent announcements on streamlining prior authorizations, issued by the U.S. Department of Health and Human Services, are a step in the right direction, but they highlight a deeper issue: Communication between payers and providers remains essentially broken. Fax machines and PDFs shouldn’t be the standard for 21st-century medicine.

AI could dramatically simplify these interactions, but only if it has access to clean, connected, context-rich data.

So, how do we fix it?

The good news: There are proven, pragmatic ways to unify data without rebuilding everything from scratch.

Modern data platforms can connect to multiple systems in real time, whether it’s your electronic health record, lab feeds, practice management tools or external payer systems. Instead of moving and copying data, these platforms create a secure, governed layer that lets AI and analytics tools access what they need, when they need it.

This reduces duplication, facilitates compliance, and gives clinicians and support staff a clearer view of each patient’s story.

And yes, it supports AI, but more importantly, it supports you.

Making the case for data products

One exciting development in this space is the rise of health care-specific data products. Think of them as curated, use-case-ready data views designed for action.

Here are a few examples already in use by leading primary care organizations:

• Patient risk summary: A unified, real-time profile combining electronic health record data, claims history and SDOH information, to highlight gaps in care or high-risk flags.
• Referral snapshot: Pulls referral status, specialty availability and previous notes into a single pane for quick follow-up.
• Prior authorization status tracker: Updates authorization requests and response statuses in real time across payer systems.

These data products aren’t just dashboards. They’re living, governed resources built with clinicians in mind, saving time, improving care decisions, and freeing up hours that would otherwise be spent searching, chasing or second-guessing.

Let’s talk about trust

Of course, none of this matters without trust. Trust in the data, trust in the AI tools and trust in the workflows that are being reshaped.

That’s why data governance is just as important as connectivity. Modern platforms enable role-based access, audit trails, consent management and zero trust architectures, so your team gets the right data without compromising privacy or security.

It’s not about giving everyone access to everything. It’s about providing the right information to the right people at the right time, securely and ethically, in real time.

From raw data to real decisions: The role of deep research and intelligent data products

To bridge the gap between overwhelming volumes of patient data and practical, frontline decision-making, new capabilities like Deep Research are proving transformational. Deep Research is a semantic query engine that enables AI systems to understand subtle context, beyond simply retrieval-based content, to deliver trusted, relevant and real-time answers from across fragmented records and systems. For primary care physicians, this means fewer blind spots and more actionable insights at the point of care. When combined with smart data products, such as a real-time medication adherence tracker, a synthetic risk scoring dashboard or a referral optimization tool, Deep Research enables these products to be populated with accurate, up-to-date information, rather than just historical data. In short, it helps physicians spend less time digging and more time caring.

It starts with data

AI won’t save American health care. But clinicians might — if we give them better tools, starting with better data. Let’s stop expecting physicians to be data janitors, AI skeptics and IT troubleshooters on top of everything else.

Let’s fix the data first, so AI can finally do what it promises.

Errol Rodericks is a vertical product marketing director at Denodo, helping health care and life sciences organizations unlock secure, trusted and real-time insights from data. He collaborates with clinicians, system integrators and platform partners to deliver pragmatic, governed data solutions that support AI, compliance and patient-centered care.