FDA Clears Tempus AI Tool to Flag Mitral Regurgitation

Tempus ECG-MRWhat to know

30-second read

For Cardiology · Internal Medicine · Family Medicine

Who

Adults 65 and older with cardiovascular risk factors or symptoms, having a standard resting 12-lead ECG collected at a healthcare facility.

What it detects

ECG patterns associated with likely undiagnosed moderate or severe mitral regurgitation; output is a binary flag, not a graded readout.

Evidence

ECG-MR’s own sensitivity, specificity and validation cohort were not published as of this story — unlike its three predecessor products, which each have reported figures.

Where it fits

Alongside a routine ECG already being read for other reasons; a positive flag is a prompt for echocardiography, not a finding to act on alone.

Key consideration

FDA’s cleared labeling states it is not a stand-alone diagnostic tool; interpret only with the ECG, other tests, symptoms and history.

Availability

Cleared 28 September 2026, Tempus’s fourth ECG-AI product. No commercial rollout date or EHR-integration partners announced.

Practice impactConsider as an echo-referral trigger

Useful only as one more opportunistic screen layered onto a test already being ordered — it changes nothing about how confirmed mitral regurgitation is managed, and its value depends entirely on validation data that has not yet been published.

Read the full physician analysis

Full analysis · 8 min readTempus AI investor release, 28 Sep 2026 · Cardiovascular Business · FDA 510(k) database (pending)

The FDA cleared Tempus ECG-MR on 28 September 2026, an AI algorithm that reads a standard, resting 12-lead ECG collected in a healthcare facility for patterns associated with moderate or severe mitral regurgitation that has not yet been diagnosed. It is the fourth ECG-AI product Tempus AI (Nasdaq: TEM) has brought through FDA clearance, following tools for arrhythmia risk, low ejection fraction and pulmonary vascular disease. FDA's own clearance language is explicit that the output — a binary flag — is not a stand-alone diagnosis and must be interpreted alongside the original ECG, other testing, symptoms and clinical history.

What the algorithm is meant to catch

Mitral regurgitation is common, often silent in its early-to-moderate stages, and diagnosed almost exclusively by echocardiography — a test that is not ordered on a patient who has no reason to raise suspicion. The premise behind ECG-MR is that a standard ECG, a test already ordered for unrelated reasons in a huge number of older adults, may carry a subtle signature of significant valve disease before a murmur is loud enough to prompt an echo referral. FDA cleared ECG-MR specifically for adults 65 and older with cardiovascular risk factors or symptoms, reading resting, non-ambulatory 12-lead ECGs collected at a healthcare facility — not a consumer wearable's single-lead trace. Tempus and Medtronic have separately run the ALERT trial, an EHR-based clinician-notification platform studying whether flagging valvular disease this way actually improves timely evaluation and treatment of aortic stenosis and mitral regurgitation; that is a distinct notification-system study, not the validation evidence behind this specific clearance, and the two should not be conflated.

The evidence gap this story cannot close

This is the section where a physician-facing story on a diagnostic clearance would normally report sensitivity, specificity and the size and design of the validation cohort. For ECG-MR, none of that was publicly available as of this writing. Tempus's own investor release and independent trade coverage (Cardiovascular Business, HLTH Insights, MPO-Mag) confirm the clearance, the date and the intended-use language, but none reports a performance figure for ECG-MR itself. That is a meaningful gap, not a minor one: Tempus's three earlier ECG-AI products each do have published numbers. ECG-AF, cleared in 2024 for atrial fibrillation risk, was validated in a peer-reviewed multicenter study in Heart Rhythm (n=4,017): 31% sensitivity, 92% specificity, 19% positive predictive value, 95% negative predictive value. ECG-Low EF, cleared in 2025 for reduced ejection fraction, had a retrospective four-site cohort study (n=14,924): 86% sensitivity, 83% specificity, 38% positive predictive value, 98% negative predictive value. Those numbers describe different algorithms detecting different conditions and say nothing about how ECG-MR performs — they are included here only to show the kind of evidence this company has published before, and that ECG-MR does not yet have a public equivalent. FDA's 510(k) summary, once posted to its public database, should settle this; it had not appeared as of this story's reporting, consistent with the two-day gap between clearance and this story.

Not a diagnosis, and not a replacement for echo

Every ECG-AI screening tool in this category, including Tempus's other three, occupies the same niche: a flag that triggers further workup, not a diagnosis that stands on its own. Echocardiography remains the diagnostic reference standard for mitral regurgitation, and nothing about a cleared screening algorithm changes that. The clinical case for a tool like this rests entirely on whether it catches real disease that would otherwise go unrecognized for longer, at a false-positive rate low enough not to flood echo labs with referrals that turn up nothing — the exact tradeoff a sensitivity/specificity figure would answer, and the one this story cannot yet answer for ECG-MR specifically.

Coding and access: unsettled

CMS assigned Category III CPT codes 0764T (concurrent) and 0765T (previously performed) to Tempus's assistive algorithmic ECG products in the “cardiac dysfunction” category, reimbursed at roughly $128.90 under APC 5734 effective 1 January 2025 — codes Tempus's own materials tie to ECG-AF and, by category, plausibly to ECG-Low EF. This reporting found no confirmation that those codes, or any billing code, extend to ECG-MR, whose indication (valvular disease) sits outside the “cardiac dysfunction” framing those codes were built around. Practices should not assume ECG-MR is billable under the existing Tempus ECG-AI codes until Tempus or CMS says so explicitly. No EHR or ECG-system integration partners, and no commercial availability date, were announced alongside the clearance.

What would change in practice

• Nothing changes for a patient with known mitral regurgitation. ECG-MR is a screening layer for the undiagnosed, not a monitoring tool for the diagnosed.
• A positive flag is a referral trigger, not a result to relay to the patient as a diagnosis. Route it to echocardiography before any conversation about valve disease.
• Do not assume reimbursement. The existing Tempus ECG-AI CPT codes were built for a different indication category; confirm coverage before relying on this as a billable service.
• Wait for the 510(k) summary before judging the tool's real-world value. Without a sensitivity, specificity or false-positive rate, there is no way to know whether this reduces missed diagnoses or mainly generates echo referrals that come back normal.

ECG-MR is a plausible extension of a screening approach Tempus has now applied to four cardiac conditions from the same routine test, and the underlying idea — that a 12-lead ECG already being collected on older, at-risk patients might carry an early signal of valve disease — is a reasonable one worth testing. What is missing, for now, is the test result: the validation data that would tell a physician whether to trust a flag from this specific product. Until FDA posts the 510(k) summary or Tempus publishes the study, this belongs in the “worth watching” category rather than the “change how I screen at-risk patients” category.

Frequently asked questions

Does a positive ECG-MR flag diagnose mitral regurgitation?

No. FDA's cleared intended use is explicit that ECG-MR is not a stand-alone diagnostic tool; its output is a binary flag meant to be read alongside the original ECG, other tests, symptoms and clinical history. Echocardiography remains the diagnostic standard for mitral regurgitation — a positive flag is a prompt to order one, not a substitute for one.

What is ECG-MR's own sensitivity and specificity?

Not published as of this story. Tempus's three earlier ECG-AI products each have their own reported validation data — ECG-AF's peer-reviewed study reported 31% sensitivity and 92% specificity for atrial fibrillation risk, and ECG-Low EF's retrospective cohort study reported 86% sensitivity and 83% specificity for reduced ejection fraction — but those numbers describe different products detecting different conditions and do not carry over to ECG-MR. Its own performance data should come from FDA's 510(k) summary once posted.

Is there a billing code for ECG-MR?

Not confirmed. CMS assigned Category III CPT codes 0764T and 0765T to Tempus's ECG-AF product for a 'cardiac dysfunction' category of assistive algorithmic ECG assessment, reimbursed around $128.90 under the Medicare physician fee schedule as of January 2025. Whether those codes, or any code, extend to ECG-MR's mitral regurgitation (valvular) indication was not confirmed in this reporting.