For decades, healthcare’s biggest paradox has been this: clinicians know what works, guidelines are well established, yet millions of patients who qualify for evidence-based care are never identified in time. Qualified Health and Anthropic are betting that large-scale, governed AI can finally close that gap.

The two companies have launched what they describe as a landmark AI deployment across the University of Texas System (UT System)—one of the largest academic health networks in the U.S.—aimed at systematically identifying patients who meet guideline-based criteria and ensuring they are evaluated for appropriate, high-quality care. The initiative brings together Qualified Health’s clinical governance platform and Anthropic’s Claude AI models, applied across vast and complex clinical datasets.

Why Evidence-Based Care Still Falls Through the Cracks

The issue isn’t a lack of medical research. Clinical guidelines and appropriateness criteria have been refined over decades. The problem is operational reality. Determining whether an individual patient meets those criteria often requires painstaking chart review across fragmented EHRs, unstructured clinician notes, lab results, imaging, and historical records.

At population scale—millions of patients and petabytes of data—this work has historically been infeasible. The consequences are significant. Tens of millions of Americans who qualify for evidence-based care are never evaluated in time. In Texas alone, an estimated 4–6 million patients fall through the cracks each year, contributing to preventable complications, higher mortality, inequities in access, and mounting pressure on already strained clinicians.

Qualified Health and Anthropic argue that this is precisely the kind of problem modern AI is suited to solve—if deployed with the right safeguards.

How the System Works

Under the new deployment, Qualified Health’s AI system—powered by Claude—continuously analyzes clinical data across the UT System. It integrates information from multiple sources, parses complex and unstructured data, and applies validated clinical guidelines and appropriateness criteria to maintain a continuously updated, population-level view of care gaps.

Rather than replacing clinical judgment, the system surfaces patients who may warrant further consideration directly into existing care team workflows. Supporting clinical context is automatically assembled, allowing clinicians to review cases efficiently and make informed decisions without wading through fragmented records.

“Healthcare is one of the most demanding environments for AI,” said Eric Kauderer-Abrams, Head of Life Sciences at Anthropic. “It requires parsing vast amounts of unstructured clinical data while operating safely within strict governance frameworks. Claude can do that reliably, and when paired with Qualified Health’s platform and a visionary health system like the UT System, it creates the conditions to deploy advanced AI safely at scale.”

From Theory to Live Clinical Use

After extensive evaluation and testing, the system is now live at the University of Texas Medical Branch (UTMB), the first deployment site within the UT System. The initial focus is cardiology, an area where delayed identification can have serious consequences.

The system evaluates unified patient profiles against precise guideline-based criteria, covering everything from guideline-directed medical therapy and medication dosing to appropriate interventional treatments for heart failure and valvular disease. Importantly, appropriateness criteria are surfaced alongside recommendations, reinforcing quality and consistency in clinical assessment.

Early results suggest the approach is resonating with clinicians:

• Complex clinical data were successfully unified into comprehensive patient profiles

• Large cohorts of previously unrecognized, high-likelihood candidates were identified

• Clinician review showed high agreement with AI-generated outputs

• Care pathways for eligible patients were accelerated

For healthcare leaders, that last point may be the most compelling. Speed matters—not just in emergencies, but in reducing the slow, systemic delays that prevent patients from ever reaching the right point of care.

Augmenting Clinicians, Not Replacing Them

Qualified Health is careful to frame the system as an augmentation tool rather than an automated decision-maker.

“The challenge isn’t that we don’t know what works,” said Justin Norden, MD, MBA, MPhil, CEO of Qualified Health. “It’s translating decades of evidence and appropriateness guidance into consistent clinical practice at scale. The system is designed to augment, not replace, clinical judgment.”

What once required extensive manual chart abstraction and coordination across systems can now happen continuously, across entire populations. In effect, the AI handles the detection and synthesis work, allowing clinicians to focus on judgment, nuance, and patient interaction.

Scaling Across Texas—and Beyond Cardiology

Building on early success at UTMB, the platform is expanding across the UT System. By the end of 2026, additional deployments are planned across primary care, vascular, gastrointestinal, rheumatology, and neurology specialties.

That expansion aligns with broader system-level goals. According to Zain Kazmi, Chief Digital & Analytics Officer and Associate Vice Chancellor of Health Affairs at the UT System, the initiative is about more than a single AI use case.

“Rather than laying solutions on top of existing systems, we are building a new shared foundation across the UT System’s health enterprise that allows new AI deployments to be introduced with consistency, accountability, and long-term impact,” Kazmi said.

The deployment is also part of the UT REAL Health AI initiative, which emphasizes two priorities: expanding access to evidence-based treatment—particularly for underserved populations—and setting a new standard for safe, responsible AI in clinical environments.

Why This Matters for Healthcare AI

As health systems nationwide evaluate population-scale AI, the UT System deployment stands out for its scope and governance-first approach. Rather than experimental pilots or narrow point solutions, this initiative aims to operationalize evidence-based medicine across entire populations.

It’s also a signal moment for Anthropic, whose Claude models are increasingly being positioned for high-stakes, regulated environments. The project has already been highlighted in Anthropic’s public communications and at industry forums such as the J.P. Morgan Healthcare Conference, underscoring growing interest in AI that can move from promise to production.

If the results continue to scale, the partnership could offer a replicable blueprint for how health systems translate clinical evidence into consistent practice—without burning out clinicians or leaving patients behind.

Get in touch with our MarTech Experts.