Cardiac-AI Explained: What Real-Time Heart Reads Change

Cardiac-AI is software that reads a heart ultrasound the way a cardiologist would — measuring how well the heart pumps — and returns an objective number in real time on the same handheld device that captured the image. That single sentence is the whole story, and as of June 2026 it stopped being a research demo and became an FDA-cleared product.

This page is the plain-English reference for the term. We explain what happened, how the mechanism works without a single equation, why it became possible now, who shipped it, the honest limits, and where small and mid-size healthcare operations should expect it to land over the next 12-36 months.

TL;DR

• On June 2, 2026, the FDA cleared Clarius Ejection Fraction AI to automatically calculate left ventricular ejection fraction (LVEF) in real time, according to PR Newswire, which describes results delivered within the 90-second window required for resuscitation guidance.

• The model was built from thousands of cardiac ultrasound images annotated by clinicians, per PR Newswire, and runs on existing wireless scanners via an app update.

• It targets primary care, emergency medicine, and rural/underserved settings — places that historically lacked sonography expertise, per DAIC.

• The point: an objective heart-function read moves from the echo lab to the exam room. The operational question for a practice is what that does to referrals, scheduling, and documentation — not whether the math is impressive.

Cardiac-AI matters because the underlying condition is common and growing. Roughly 6.7 million US adults live with heart failure today, a count projected to keep rising, according to Heart Failure Society of America. A faster, cheaper first read at the point of care touches a large population.

What actually happened

Clarius, a handheld wireless ultrasound maker, received FDA clearance on June 2, 2026 for an AI feature that computes ejection fraction automatically from standard cardiac windows. According to PR Newswire, the FDA cleared the tool on June 2, 2026 to calculate the percentage of blood the heart pumps out with each beat and returns it on the device during the scan.

Ejection fraction is the single most quoted number in cardiac function. A normal heart ejects a healthy fraction of its blood volume per beat; a struggling one ejects less. According to DAIC, the tool flags EF declining below 40%, indicating possible heart failure — the threshold clinicians use to separate reduced from preserved function.

The clearance is notable not because EF is new, but because of where it now gets measured. Historically, quantifying cardiac function "required a high level of sonography expertise or a wait for formal echocardiography," in the words of Sarah Leverett, Clarius VP of Marketing, quoted by PR Newswire. The product collapses that into a real-time read on a handheld scanner.

The mechanism, in plain language

There are no equations here — just three steps.

First, capture. A clinician points the scanner at one of two standard views: the parasternal long-axis or the apical 4-chamber. Per DAIC, those are the two windows the tool reads, and an early tester described receiving "a calculated EF just from a parasternal long or 4 chamber view" as "magical."

Second, recognize. The AI was trained on thousands of clinician-labeled images, so it has learned what a left ventricle looks like across body types and image quality. It traces the chamber's borders as the heart fills and empties — the same mental task a sonographer performs, automated.

Third, report. It compares the filled and emptied chamber and returns the ejection-fraction percentage on screen, removing the subjective "eyeball estimate" that varies between operators. Per PR Newswire, it works "across varying patient body types and image qualities" with "minimal workflow steps from imaging to result."

Why now — what constraint broke

Three things had to line up, and as of June 2026 they did.

The first was hardware. Diagnostic-grade ultrasound used to mean a cart in a hospital. Clarius scanners are wireless handhelds, and per PR Newswire, nearly 7 million high-definition scans have already been performed on the platform across 70+ countries — enough deployed devices to make a software feature worth shipping.

The second was data. Real-time inference of a clinical measurement needs a large, clinician-labeled training set. The model was developed from thousands of annotated cardiac images, per DAIC — the labeled corpus that older eyeball-estimate workflows never assembled.

The third was cost pressure. Formal echocardiography is expensive and varies wildly: UnitedHealth found the price paid for an echocardiogram ranged from $210 to $1,830 across 12.5 million diagnostic tests, per Healthcare Dive. When a first-pass read can happen in the exam room, the economics of who orders what change.

Sources for the table above: scan and country counts per PR Newswire; labeled-image corpus per DAIC; echo price range and test volume per Healthcare Dive.

How the old workflow compared

To see why a real-time read matters, it helps to lay the old path next to the new one. The conventional route to an ejection fraction was: order a formal echocardiogram, schedule the patient, have a trained sonographer acquire a full study, and wait for a cardiologist to read and report it — often days later, and at a price that, per Healthcare Dive, varied from $210 to $1,830 per test. The new path collapses the first, screening-grade read into the same encounter.

That comparison is the whole reason the term "cardiac-AI" is worth a page. The 90-second figure and the $210–$1,830 range are both sourced — to PR Newswire and Healthcare Dive respectively — and together they explain the pull: faster, cheaper, and usable by clinicians who aren't echo specialists.

It's worth being precise about what "screening grade" means. The point-of-care read does not replace the full study; it tells you whether you likely need one. In a world where formal echo capacity is finite and expensive, a fast filter at intake is operationally valuable even if it never produces a final diagnosis on its own.

The clinical context that makes this matter

Cardiac-AI is not interesting in a vacuum — it's interesting because heart disease is common and heart failure is rising. Heart disease prevalence among US adults aged 18 and over was 5.5% in 2018, according to CDC, and it climbs steeply with age: among adults 75 and older it reached 24.2% in 2019, according to CDC. A screening tool that works in primary care meets that population where it already shows up.

Heart failure specifically is both prevalent and growing. Roughly 6.7 million US adults have heart failure today, with prevalence rising as the population ages, according to Heart Failure Society of America. The single threshold cardiac-AI flags — an EF below 40% per DAIC — is precisely the line clinicians use to separate reduced-ejection-fraction heart failure from preserved-ejection-fraction cases, which is why an automated read of that number is more than a convenience.

Sources: heart-disease prevalence (2018, 2019) per the CDC; heart-failure count per the Heart Failure Society of America; EF threshold per DAIC.

Who shipped it

Clarius shipped the feature; the FDA cleared it. Per PR Newswire, it runs on the Clarius PA, PAL, and C3 HD3 wireless scanners via an app update, and access requires a Clarius Membership or one-time license. There is no new hardware to buy if a practice already owns a compatible scanner — it is a software entitlement.

The clinical framing came from the people who tested it. Dr. Brian Johnson, an emergency physician and early tester, called the ability to get a calculated EF from a single standard view "magical," per PR Newswire. That is the use case: a clinician who is not a cardiac sonographer getting a defensible number fast.

The honest limits

A hub page that only sells the upside is useless. Here is what cardiac-AI does not do.

It does not replace a formal echocardiogram or a cardiologist's read. It produces one number — EF — from two views. A complete echo measures valves, wall motion, diastolic function, and chamber sizes. An objective EF is a triage and screening signal, not a diagnosis.

It also inherits the limits of the operator's image. The tool is robust across body types per the vendor, but a poor acoustic window still produces a poor read. And it is gated behind a paid Clarius entitlement on specific scanners, so it is not a free upgrade for every handheld in the field.

Finally, an EF number creates downstream work. A flagged EF below 40% (DAIC) means a referral, a follow-up, documentation, and often a confirmatory echo. The measurement is fast; the operational tail is not — which is exactly where workflow automation enters the story.

What "real-time" actually buys you

The phrase "real-time" gets thrown around loosely, so it's worth grounding. The concrete claim is that results arrive within the 90-second window required for resuscitation guidance, per PR Newswire. That number is not marketing — it maps to a real clinical constraint. In a resuscitation, decisions happen in seconds, and a number that arrives after the patient has been stabilized (or not) is useless. A read that lands inside that window can change what the team does next.

Per DAIC, the tool is positioned to help evaluate undifferentiated shortness of breath, septic shock, and trauma — exactly the presentations where a clinician needs to know quickly whether the heart is pumping adequately. The speed isn't a luxury feature; it's what makes the read clinically actionable in the settings the vendor targets.

There's a quieter operational benefit too. Because the read is objective rather than an "eyeball estimate," two clinicians scanning the same patient should land near the same number. That consistency is what lets a downstream system trust the value enough to act on it automatically — you can't build a reliable referral trigger on a subjective guess that varies by who held the probe.

Where the workflow automation fits

A real-time EF is the trigger event, not the finish line. The moment a scanner returns a low number, a practice needs to: document the result in the EHR, route a referral to cardiology, schedule the confirmatory echo, verify insurance, and follow up. Today most of that is manual.

This is the seam where automation belongs. Teams that already route clinical documents, intake, and referrals through US Tech Automations workflows can treat a flagged EF as a structured trigger that opens a referral task and a scheduling request automatically — the read fires the workflow rather than landing in an inbox. The point isn't the AI read itself; it's that the read becomes a clean event other systems can act on.

For practices building this out, the design question is mundane and important: which EF threshold opens a referral, who approves it, and how the confirmatory echo gets booked. An automated read makes a natural automation trigger, and according to DAIC an EF below 40% flags possible heart failure — but a human still owns the clinical decision. This approval-gated routing is the pattern US Tech Automations builds for clinical operations teams. We cover the practice-level mechanics in what cardiac-AI means for healthcare practices.

Signal vs Speculation

Everything above is sourced fact. This section is our forecast — read it as opinion.

Our read: if point-of-care EF reads become routine in primary care and emergency settings, the bottleneck shifts from measurement to coordination. The scarce resource stops being "who can run the scan" and becomes "who handles the referral and confirmatory echo when a number comes back abnormal." Given heart-failure prevalence near 6.7 million adults and rising per the Heart Failure Society of America, more first-pass reads plausibly means more downstream referrals — a volume problem, not a clinical one.

Our read: the cost story is the quiet driver. With echocardiogram prices spanning $210 to $1,830 per Healthcare Dive, payers have every incentive to favor a cheaper first read that filters who actually needs the expensive confirmatory study. We expect reimbursement and protocol pressure to push point-of-care EF into intake within 12-36 months, especially in rural settings the vendor explicitly targets.

Our read: for small and mid-size practices, the winners over the next 1-3 years will be the ones who treat the EF read as a workflow trigger from day one — wiring documentation, referral routing, and scheduling around it — rather than bolting on a fancy scanner and leaving the coordination manual. None of this is guaranteed; it is our interpretation of where the facts point.

Key Takeaways

• Cardiac-AI is software that returns an objective ejection-fraction number in real time at the point of care, FDA-cleared June 2, 2026, per PR Newswire.

• It works from two standard views and flags EF below 40% as possible heart failure, per DAIC.

• It became possible because deployed handheld scanners (~7M scans), labeled training data, and echo-cost pressure ($210–$1,830 per test) lined up at once.

• It is a triage signal, not a replacement for formal echo or a cardiologist.

• The operational value lives downstream: documentation, referral routing, and scheduling — the parts worth automating.

Cardiac-AI turns a heart read into a fast, objective event. Capturing the value means wiring that event into the systems that handle what comes next — and that is a workflow-design problem teams can solve today with agentic automation workflows. If you run a clinic, start with the practice-level guide on what cardiac-AI means for healthcare practices.

Frequently Asked Questions

What is cardiac-AI?

Cardiac-AI is software that automatically reads a heart ultrasound and returns a clinical measurement — most prominently ejection fraction — in real time. The first FDA-cleared example, Clarius Ejection Fraction AI, was cleared June 2, 2026, per PR Newswire.

What is ejection fraction and why does it matter?

Ejection fraction is the percentage of blood the heart pumps out with each beat. It is the most common measure of heart-pump function, and a value below 40% flags possible heart failure, per DAIC.

Does cardiac-AI replace a cardiologist or a full echocardiogram?

No. It produces one objective number from two standard views as a triage and screening signal. A complete echocardiogram and a specialist read remain necessary for diagnosis, and the tool is positioned as a first-pass read, per PR Newswire.

How fast is a real-time ejection-fraction read?

Fast enough for emergency use: results arrive within the 90-second window required for resuscitation guidance, per PR Newswire.

Why is cardiac-AI emerging now?

Because deployed handheld scanners (nearly 7 million scans across 70+ countries), a clinician-labeled training corpus, and the high, variable cost of formal echocardiography ($210–$1,830 per test) converged, per PR Newswire and Healthcare Dive.

How does cardiac-AI affect a practice's workflow?

The read itself is fast; the work is downstream. A flagged result needs documentation, a cardiology referral, a confirmatory echo, and follow-up — the coordination layer that practices can automate by treating the EF result as a structured trigger.

Freshness note: written as of June 2026, based on the June 2, 2026 FDA clearance.