Doxiverse Staff, August 2026
The American Heart Association's AI Assessment Lab has published its first known evaluation of a commercial cardiovascular AI tool, finding that Ultromics' FDA-cleared EchoGo Heart Failure algorithm could identify a commonly missed form of heart failure an average of 263 days earlier than standard clinical care. The report, based on a modeled analysis of real-world data, was announced by Ultromics on July 22, 2026.
Why it matters: This is a case of a major cardiology organization running third-party validation of a vendor's AI diagnostic using outside data rather than the company's own studies which gives hospitals evaluating similar tools an independent reference point instead of relying solely on vendor-supplied evidence.
What Happened
The AHA's AI Assessment Lab, an initiative the AHA says provides scientifically rigorous evaluations of cardiovascular and stroke AI algorithms. The report evaluated EchoGo Heart Failure, an AI tool from UK-based Ultromics that analyzes standard echocardiogram videos to help identify heart failure with preserved ejection fraction (HFpEF).
HFpEF is among the most common forms of heart failure, but its symptoms overlap with other conditions and, the diagnosis is particularly difficult among women and people of color, who often face delays in identification and treatment.
The analysis was based on an independent dataset curated by Dandelion Health, a real-world data and clinical AI platform, rather than data supplied directly by Ultromics. Using this dataset, the report modeled outcomes on a per-10,000-patient basis over a five-year period. Key modeled findings, as reported by Ultromics, include:
- EchoGo Heart Failure could have identified HFpEF an average of 263 days (roughly nine months) earlier than standard diagnostic practice among patients who would otherwise have faced delayed diagnosis.
- Earlier detection was projected to improve survival outcomes and reduce healthcare utilization. Ultromics' release states this could translate to nearly 500 additional lives saved per 10,000 patients, though the underlying calculation was not detailed in the materials we reviewed.
- The report projected up to approximately $1.9 million in additional revenue over five years for health systems, with savings of roughly $1,800 per patient from both health-system and payer perspectives.
Roger Owens, Ultromics' chief commercial officer, told that "diagnostic accuracy is what earns the AI a seat at the table." In a separate statement to Diagnostic and Interventional Cardiology, Owens said AI diagnostics face a higher evidentiary bar than traditional cardiovascular technologies precisely because they're newer, and argued that hospitals need clear evidence of how these models are trained, validated, and integrated into clinical workflows before they'll trust them in practice.
What This Means for Cardiologists and Hospital Administrators
For clinicians, the report adds independently sourced though modeled, not directly observed evidence that an already FDA-cleared echocardiogram AI tool may catch HFpEF cases that would otherwise be missed or delayed. That could be useful context when evaluating whether to adopt or expand use of the tool in an echo lab.
For hospital administrators weighing AI diagnostic purchases more broadly, the more significant development may be structural: the AHA now appears to have a standing mechanism for third-party review of cardiovascular AI claims. That could become a resource to consult when vetting other vendors' cardiovascular AI tools going forward, rather than relying solely on FDA clearance status or vendor-published data.
What's Still Unsettled
- This is a modeled projection, not a prospective clinical trial. The lives-saved, revenue, and diagnosis-count figures come from a retrospective, per-10,000-patient model applied to a curated dataset, not from tracking real patient outcomes after EchoGo Heart Failure was deployed in specific health systems.
- The report was announced via the vendor. Every specific figure in this piece traces back to Ultromics' own press release and coverage built on it. We could not locate the full underlying AHA Impact Report document to independently verify the methodology behind the "263 days," "$1.9 million," or "4,173 diagnoses" figures.
- "Independent" needs context. The dataset and analysis came through the AHA's AI Assessment Lab and Dandelion Health rather than from Ultromics directly, which is a meaningful step toward independence but the report was still publicized as part of Ultromics' own marketing, and the AHA's specific role in reviewing or approving the cited figures wasn't detailed in our source material.
Sources
- Ultromics, press release, July 22, 2026: "American Heart Association AI Assessment Lab Report Finds Ultromics' EchoGo Heart Failure May Identify HFpEF Earlier Than Standard Care" (primary vendor source)https://www.ultromics.com/press-releases/aha-ai-assessment-lab-report-finds-ultromics-echogo-heart-failure-may-identify-hfpef-earlier-than-standard-care
- Medical Device Network, "New AHA report highlights Ultromics' AI tool's benefit in earlier heart failure diagnosis," undated but published on or shortly after July 22, 2026.