Beyond the AI Hype: How Hybrid Intelligence is Reshaping Clinical Data

SAN FRANCISCO, CA – June 16, 2026 – Carta Healthcare announced today that its Lighthouse platform has expanded to support an additional 21 clinical registries, bringing its total to over 70. While any expansion in the health-tech space is noteworthy, the real story isn't the number, but the methodology driving it. The company’s growth is built on a model it calls “Hybrid Intelligence,” a pragmatic fusion of artificial intelligence and human expertise that offers a grounded counter-narrative to the hype of full automation.

In an industry grappling with how to responsibly deploy AI, this approach, which keeps a “clinician at the helm,” is gaining significant traction. By focusing on augmenting rather than replacing skilled professionals, it delivers quantifiable improvements in efficiency and accuracy. This shift from rhetoric to results provides a crucial blueprint for how innovation can meet execution in the high-stakes world of healthcare.

The High Stakes of Clinical Data

To understand the significance of this model, one must first appreciate the complex world it operates in. Clinical registries—vast, curated databases of patient data for specific diseases, conditions, or procedures—are the bedrock of modern quality improvement. Organizations like the American College of Cardiology (ACC) and the Society of Thoracic Surgeons (STS) use them to set benchmarks, track outcomes, and drive life-saving advancements.

However, contributing to these registries is a monumental task. The process, known as clinical data abstraction, involves highly trained specialists, often nurses, manually combing through patient records to find and codify hundreds of specific data points. It’s a painstaking process made more difficult by the fact that up to 80% of a patient's story is locked away in unstructured text like physician's notes and discharge summaries. Each registry has its own unique language, with distinct definitions and submission requirements.

“The clinical logic behind a cardiac procedure case is nothing like the logic behind a trauma or neonatal case,” explained Betsy Castillo, VP of Clinical Data Abstraction at Carta Healthcare. This complexity creates a significant operational bottleneck for hospitals, consuming thousands of hours of skilled labor and diverting clinical staff from direct patient care. The result is often a trade-off between speed, cost, and data quality—a compromise healthcare systems can ill afford.

When 'Good AI' Isn't Good Enough

Into this environment enters artificial intelligence, promising a revolution in efficiency. Yet, early attempts at pure automation have revealed the limitations of general-purpose AI. While a standard AI model can be trained to scan documents for keywords, it lacks the nuanced clinical judgment to interpret context correctly. It might identify a condition mentioned in a patient’s family history and mistakenly code it as a current diagnosis, or miss the subtle distinction between a suspected and a confirmed condition.

As Castillo notes, “When AI is built without that understanding, it still produces answers. They just miss the clinical context.” These seemingly small errors have massive downstream effects, leading to failed data submissions, skewed quality metrics, and flawed clinical insights. Recent industry surveys underscore the professional skepticism toward unchecked AI; one from March 2026 found that 74% of healthcare professionals identify the misinterpretation of complex clinical data as a top risk of running AI without human oversight. An overwhelming 97% believe AI should support, not replace, clinical expertise.

“Every registry represents a distinct clinical discipline with its own quality metrics and accuracy thresholds,” said Brent Dover, CEO of Carta Healthcare. “Building AI that works for one registry does not mean it works for another. Every registry has unique complexity, and supporting each one at the level health systems need requires rigor, investment, and care.” This is the core challenge that purely automated solutions have struggled to overcome.

Executing on Hybrid Intelligence

Carta Healthcare’s Hybrid Intelligence model is engineered as a direct response to this challenge. Instead of aiming to remove the human, it empowers them. The platform’s AI acts as a force multiplier, performing the initial, time-consuming work of scanning vast electronic medical records and surfacing suggested answers for registry questions. The crucial step, however, is that these AI-generated findings are then presented to a credentialed clinical data abstractor for validation.

This keeps an expert at the center of the process, transforming their role from a manual data hunter to a high-level reviewer. The abstractor’s deep domain expertise is used to verify the AI's findings, correct any misinterpretations, and handle the ambiguous edge cases where human judgment is irreplaceable. As Castillo puts it, “That's where expert abstractors at the helm, continually verifying and improving the AI, change everything.”

The quantifiable results of this model are compelling. According to the company, health systems using this approach have reduced abstraction time by up to 66% and cut associated costs by 50% or more. Critically, this efficiency does not come at the expense of quality. The model consistently achieves Inter-Rater Reliability (IRR) scores—a measure of data accuracy—greater than 98%. This combination of speed and precision, backed by a 100% customer retention rate, demonstrates a product that is not just innovative in theory, but effective in practice.

From Data Quality to Strategic Advantage

The benefits of this refined process extend far beyond the data abstraction department. For hospital executives, the cost and time savings translate into a direct strategic advantage. Resources previously tied up in manual data entry can be reallocated to patient-facing activities and other quality improvement initiatives.

More importantly, the delivery of consistently accurate and timely data transforms an administrative burden into a powerful asset. One 10-hospital system, for example, used the enhanced data completeness from this model to improve its vascular quality initiative star ratings from mostly one and two stars to three stars across all its sites. Such improvements have a direct impact on a hospital's reputation, reimbursement rates, and ability to attract top talent.

High-quality registry data enables clinical leaders to identify trends, benchmark performance against national standards, and make more informed decisions about patient care protocols. When data is reliable, it can be trusted to guide everything from post-discharge care plans aimed at reducing readmissions to system-wide safety initiatives. In the end, better data facilitates better medicine.

The industry appears to be moving past the initial, simplistic narrative of AI as a job-killer. The success of hybrid models suggests a more mature understanding is taking hold: that the true power of this technology lies in its ability to amplify human intelligence. By focusing on execution over hype and pairing advanced technology with irreplaceable human expertise, this approach provides a clear and effective path toward a more efficient and intelligent healthcare system.