FDA clears Cardiosense software to estimate a heart-failure filling pressure without a catheter

The Food and Drug Administration has granted a De Novo classification to Cardiosense’s PCWP Analysis Software, creating a new regulatory category for a device that estimates a key heart-failure filling pressure without threading a catheter into the heart. The agency’s openFDA record lists the decision as DEN250057, a Direct De Novo for Chicago-based Cardiosense, Inc., with a decision date of May 22, 2026; the record’s review advisory committee field carries the cardiovascular review-panel code “CV.”

The pressure in question is pulmonary capillary wedge pressure (PCWP), a measure of how congested a failing heart is. Today it is captured either by right heart catheterization or by an implanted pulmonary-artery sensor — both invasive. The software instead reads signals from a wearable chest sensor and infers the value.

The authorized indication is narrow: adults with heart failure with reduced ejection fraction (HFrEF), defined as a left ventricular ejection fraction of 40% or less, who have New York Heart Association (NYHA) functional class II, III, or IV symptoms.

What the evidence shows

The authorization rests on SEISMIC-HF I, a multicenter prospective study published in JACC: Heart Failure. Investigators enrolled 310 patients with HFrEF (EF ≤40%) across inpatient and outpatient settings (DOI). The wearable — Cardiosense’s CardioTag — records electrocardiography, seismocardiography, and photoplethysmography, and a machine-learning model maps those signals to an estimated PCWP.

The reference standard was right heart catheterization, with a blinded core laboratory adjudicating the tracings to set the criterion-standard labels. The model was then tested on a held-out set unseen until final evaluation.

In that test set, estimated PCWP differed from catheter-measured PCWP by 1.04 ± 5.57 mm Hg, with limits of agreement of −9.9 to 11.9 mm Hg.

The cohort’s mean measured PCWP was 18.1 ± 9.45 mm Hg. Patients averaged 61 ± 13 years, were 38% female, and were 44% White and 39% African American; the authors described performance as consistent across sex, race, ethnicity, and body-mass index (DOI).

Two caveats temper the result. The wide limits of agreement — roughly ±11 mm Hg — mean an individual reading can land well off the catheter value, and the authorized use is confined to HFrEF (EF ≤40%) patients with NYHA class II–IV symptoms; the larger preserved-EF population was not studied here. Whether a non-invasive estimate translates into the hospitalization reductions seen with implanted sensors is a separate, unanswered question. The De Novo establishes a device class; it is not evidence that monitoring changes outcomes.