We think the honest thing is to show our work while it is still in progress. Here is how PulSentry is being validated, what stage each part is at, and what we are careful not to claim yet.

Status in one line: PulSentry is investigational and not FDA-cleared. Our validation is pre-publication: results are not yet peer-reviewed, and nothing here should be read as a performance claim.

A two-part validation strategy

We are validating the pulsus index in two complementary ways: a retrospective analysis on a large library of existing physiologic recordings, and a prospective observational study in living clinical workflows.

1. Retrospective: a MIMIC-IV waveform subset

To characterize how the pulsus index behaves across a wide range of physiologic states, our collaborators are analyzing a subset of the MIMIC-IV Waveform Database, a widely used critical-care research resource from the MIT Laboratory for Computational Physiology. The working subset is roughly 100 patients comprising several thousand hours of waveform data, chosen to include the high end of the severity spectrum. MIMIC-IV is a credentialed-access resource; access requires individual training and a signed data use agreement through PhysioNet, and we do not redistribute any patient-level data. This arm is retrospective and hypothesis-generating.

2. Prospective: an observational study at Harbor-UCLA

In parallel, a prospective observational study captures the PPG waveform in clinical workflows, beginning with emergency-department patients who have shortness of breath without tamponade (the lower end of the spectrum). A planned addition would enroll patients referred for echocardiography, so that the echo result serves as the reference standard. The prospective work is being conducted under institutional review.

What the reference standard is for each arm

For the tamponade question, the reference standard is echocardiography. For the broader breathlessness question, the reference is clinical adjudication. The retrospective MIMIC-IV arm is for characterization and hypothesis generation, not for establishing diagnostic performance.

What we are careful not to claim

We do not claim a validated sensitivity or specificity for any use. That awaits completed, peer-reviewed analysis.
We do not claim FDA clearance, and we do not claim a dyspnea indication.
We describe a signal (respiratory variation in the PPG waveform), not a diagnosis.
When we publish, we will say plainly whether results are peer-reviewed or a preprint.

How this connects to the rest of the site

For the clinical rationale, see why echo misses post-surgical hematoma. For the research-stage breathlessness direction, see the respiratory-signal review. For a plain-language overview, see the patient and family guide.

How we are validating PulSentry