Tempus Cleared an ECG Model, But Published No AUC, Sensitivity Or Cohort

Tempus AI announced FDA 510(k) clearance on Monday for Tempus ECG-PH, software that reads a standard resting 12-lead ECG and emits a binary flag for signs of mean pulmonary artery pressure above 20 mmHg. Intended population: patients 40 or older presenting with dyspnea, fatigue, chest pain or edema, with no known history of pulmonary hypertension. It is the company's third cleared cardiovascular device.

What the conventional framing gets wrong

Several outlets ran "FDA Approves." It did not. A 510(k) clearance is a finding of substantial equivalence to an already-cleared predicate device; a PMA approval is a determination of safety and effectiveness on the device's own evidence. They are different regulatory acts with different evidentiary bars, and the words are not interchangeable. The second and more consequential problem: the release publishes no AUC, no sensitivity, no specificity, no cohort size and no study design. Every performance figure now circulating about this device has been imported from somewhere other than the announcement. Third, the device does not diagnose pulmonary hypertension. It emits one bit.

The limitations, in the company's own words

"Tempus ECG-PH only analyzes ECG data and provides a binary output for interpretation." It is "not intended to be a stand-alone diagnostic tool for PH," "should not be used for serial patient monitoring," and "should not be used on ECGs with paced rhythms." Those constraints are printed in the release and are largely absent from the coverage.

Why the underlying idea is still significant

Pulmonary hypertension affects roughly 1% of the population and up to 10% of adults over 65, and confirming it today requires right-heart catheterisation — an invasive procedure nobody orders speculatively. A screening flag derived from an ECG that has already been taken would move a great many referrals. That is precisely why the absence of any published performance figure matters: a screening test's entire value is its sensitivity and its false positive rate, and neither has been disclosed.

How the market read it

Tempus shares fell 8.97% intraday. No K-number for the device had appeared in the FDA's public 510(k) database at the time of writing, so the company's release remains the only primary record.