Cardiology · Noninvasive haemodynamic estimation (wearable sensor + software)
PCWP Analysis Software
Cardiosense, Inc.
The first software the FDA has classified for estimating a cardiac filling pressure from a wearable sensor. Its pivotal study is a prospective, 15-site validation against right heart catheterisation in 310 HFrEF patients. Only 233 recordings were usable, and the headline figures come from a held-out fifth of those, about 47. Mean error was about 1 mmHg, but the limits of agreement were roughly plus or minus 11 mmHg. Four months on the US market, no outcome trial, and no record yet in any register outside the United States.
Performance Metrics
Clinical Evidence
The evidence is one prospective validation study of the algorithm against the invasive reference, plus the small feasibility work that preceded it. SEISMIC-HF I (Klein et al., JACC: Heart Failure 2025) enrolled patients already scheduled for a right heart catheterisation (RHC) at 15 US sites and fitted them with the CardioTag sensor during the procedure. The published analysis covers 310 patients with HFrEF (ejection fraction 40% or below). A blinded core laboratory adjudicated the RHC wedge tracings. The data were split into a training set and a held-out test set unseen until the final evaluation. In the test set the model estimated PCWP with an error of 1.04 plus or minus 5.57 mmHg and limits of agreement of -9.9 to 11.9 mmHg. The authors report consistent performance across sex, race, ethnicity and body mass index. An open-access summary of the AHA 2024 presentation by two site investigators (Cheema and Tibrewala, Heart Failure Reviews 2025) adds what the abstract omits: 943 patients were enrolled in the parent study, 233 of the 310 HFrEF recordings passed quality control, the split was 80/20, and the held-out correlation with RHC was r = 0.74 (R squared 0.55). So the error and agreement figures above rest on a test set of roughly 47 recordings, not 310 patients. They note the subgroup sample sizes did not allow rigorous statistical testing. Three things the evidence does not yet show. First, no readable source reports sensitivity and specificity for the cleared 18 mmHg threshold; the FDA decision summary was not posted at the time of review and the JACC full text could not be read, so the classification performance is unstated here rather than absent. Second, all measurements were taken in the catheterisation laboratory. Performance at home, the setting the technology is aimed at, is untested. Third, there is no trial of management guided by the estimate, so nothing yet links the output to hospitalisation or mortality. The 2022 feasibility paper (Shandhi et al., IEEE Transactions on Biomedical Engineering) is earlier research on 20 patients with a wearable seismocardiography patch during vasodilator challenge; it tracked changes in PCWP, not absolute values, and is background rather than evidence for the cleared device. The pivotal study was funded by Cardiosense. Of the 21 authors, three are company employees and one is a company consultant. The senior author holds equity in the company and sits on its board, and one more author reports a financial interest in it. The first author consults for Cardiosense and for Abbott, Edwards and Medtronic. A second author lists Cardiosense among consulting or ownership ties, and two more report Cardiosense research grants. AurisMark: one mark. The device holds one regulatory authorisation and has prospective published evidence, but it has been on the market for four months with no documented uptake, so the two-mark bar of broad regional adoption is not met.
| Study | Design | n | Sensitivity | Specificity | AUC | Published |
|---|---|---|---|---|---|---|
| Klein L, Fudim M, Etemadi M, et al. (Inan OT, senior) | ProspectiveProspective | 310 | — | — | — | JACC: Heart Failure, 2025 (PMID 40542792); SEISMIC-HF I, multicentre prospective validation against core-lab-adjudicated RHC in HFrEF; 310 enrolled, 233 passed quality control, 80/20 split; held-out error 1.04 ± 5.57 mmHg, limits of agreement -9.9 to 11.9 mmHg; consistent across sex, race, ethnicity and BMI |
| Shandhi MMH, Fan J, Heller JA, Etemadi M, Klein L, Inan OT | ProspectiveProspective | 20 | — | — | — | IEEE Transactions on Biomedical Engineering, 2022 (PMID 35100106); single-centre feasibility study of a wearable seismocardiography patch during vasodilator challenge at RHC; estimated changes in PCWP with RMSE 1.9 mmHg (training-testing, n=15) and 2.9 mmHg (validation, n=5); research precursor, not the cleared device |
Inside the algorithm
Editorial featureHow PCWP Analysis Software estimates a filling pressure.
Five stages — from raw input to verdict — drawn from manufacturer documentation and the public regulatory record.
- INGEST
- NORMALISE
- DETECT
- LOCALISE
- VERDICT
Stage 01 · INGEST
Three signals from one chest-worn patch.
The input is a recording from CardioTag, a wearable sensor placed on the sternum. It captures three synchronised signals: the seismocardiogram (chest-wall vibrations from each heartbeat), the electrocardiogram (electrical timing) and the photoplethysmogram (pulse volume). In the pivotal study the recording was taken while the patient lay on the catheterisation table, so the reference wedge pressure and the wearable signals were captured at the same time.
Input
Multimodal wearable recording
Inference
Per-recording, adjunctive
Inside the Auris+ Listing
Five more sections complete this device’s Auris+ Listing.
Decision Ledger
ProPro unlocks a private, cross-vendor log of every case you read on this device — what the AI called, what you concluded, and a one-line reason if you overrode it. Ready for the EU AI Act's deployer logging obligations when they land in 2028.
Clinical Evidence Deep Dive
ProPro unlocks the structured clinical-evidence summary — study count, target patient population, and a tabular accuracy-metrics view drawn from peer-reviewed sources.
Peer-Reviewed Publications
ProPro unlocks the curated peer-reviewed publication list with PubMed cross-links — the citation backbone of every editorial verdict.
Post-Market & Regulatory Conditions
ProPro unlocks the post-market surveillance summary, recall record, and the conditions of approval that bound real-world use.
AI Algorithm Version History
ProPro unlocks the chronological record of algorithm version changes — what changed when, drawn from manufacturer changelogs and regulatory filings.
Regulatory Approvals
Safety Record
No adverse event reports, recalls or enforcement actions naming Cardiosense were found in the openFDA MAUDE, recall and enforcement databases on 29 September 2026, four months after the grant. The classification order names the risks FDA identified for the device type: delayed or incorrect treatment from misinterpretation or overreliance, inaccurate pressure estimation, and absent or erroneous output from software error. The mitigations are clinical performance testing, usability assessment, labelling that carries the performance figures with confidence intervals, and a mandatory post-market performance management plan. The pivotal publication's abstract has no safety section, so no statement about adverse events in the study can be made here.
Intended Use & Indications
From the De Novo classification order (DEN250057, 22 May 2026): PCWP Analysis Software is intended to noninvasively estimate pulmonary capillary wedge pressure (PCWP) and identify patients with a PCWP above or below 18 mmHg, for patients diagnosed with heart failure with reduced ejection fraction (HFrEF) and New York Heart Association (NYHA) Functional Class II, III, IV symptoms. The hemodynamic data will be used by qualified healthcare professionals in conjunction with other standard of care parameters to evaluate heart failure patients. The results are intended to be used under the supervision of a qualified healthcare professional in conjunction with the patient's clinical history, symptoms, and other diagnostic tests, as well as the professional's judgment. Patient management decisions should not be made solely on the results of the software. Prescription use only.