Cardiology · Automated echocardiography analysis & reporting
Us2.ai
Eko.ai Pte. Ltd. (Us2.ai)
An automated second pair of hands for the echo lab. Us2.ai turns a full transthoracic study into a structured, reproducible report in seconds, and its measurements have held up in prospective validation against expert core labs and, notably, in the hands of two-week-trained novices with handheld probes. The evidence is measurement accuracy and workflow, not a randomised outcome trial — a distinction worth keeping in view.
Performance Metrics
Clinical Evidence
Us2.ai's evidence base is about measurement agreement and screening accuracy, and it reads best when kept in that frame. The founding work is a multicohort study in Lancet Digital Health (2022) that developed and tested the fully automated workflow for classifying, segmenting and measuring 2D and Doppler echocardiography. A companion formal-validation study in Nature Communications (2022) compared the deep-learning interpretation of 23 parameters against three repeated measurements by core-lab sonographers across 602 studies, and found the disagreement between the algorithm and the humans no larger than the disagreement among the human readers themselves — the relevant bar for an automated measurement tool. Two prospective studies extend the case toward the point of care. PANES-HF (Scientific Reports, 2024) had laypeople with no prior echo experience acquire handheld images after two weeks of training, with Us2.ai reporting: the AI-enabled pathway reached an area under the curve of 0.88 for detecting reduced ejection fraction, with sensitivity 84.6 per cent and specificity 91.4 per cent for a left-ventricular ejection fraction below 50 per cent. The larger real-world OPERA analysis (European Journal of Heart Failure, 2025) applied the software to handheld studies in 867 patients with suspected heart failure and reported diagnostic accuracy of 0.93 for identifying an ejection fraction of 40 per cent or below, interchangeable with cart-based readings by two experts. What none of these studies is: a randomised trial of patient outcomes. The device's proven contribution is faster, reproducible measurement and a workable novice-acquisition pathway, not a demonstrated change in mortality or management endpoints.
| Study | Design | n | Sensitivity | Specificity | AUC | Published |
|---|---|---|---|---|---|---|
| Tromp J, Seekings PJ, Hung C-L, et al. | ProspectiveProspective | 1,551 | — | — | — | Lancet Digital Health, 2022 (PMID 34863649); multicohort development and testing of a fully automated deep-learning workflow for 2D and Doppler echocardiogram interpretation (1145 training, 406 internal test) |
| Tromp J, Bauer D, Claggett BL, et al. (Solomon SD, corresponding) | ProspectiveProspective | 602 | — | — | — | Nature Communications, 2022 (PMID 36351912); formal validation of 23 automated parameters vs three repeated core-lab sonographer measurements — algorithm–human disagreement no larger than inter-human disagreement |
| PANES-HF investigators (point-of-care AI-enhanced novice echocardiography) | ProspectiveProspective | 100 | 84.6% | 91.4% | — | Scientific Reports, 2024 (PMID 38866831); novice handheld acquisition after 2 weeks' training with Us2.ai reporting; AUC 0.88 for reduced LVEF, sensitivity/specificity for LVEF <50% |
| Campbell A, et al. (OPERA) | ProspectiveProspective | 867 | — | — | — | European Journal of Heart Failure, 2025 (PMID 40702880); real-world handheld echo in suspected heart failure, automated LVEF accuracy 0.93 for LVEF ≤40%, interchangeable with cart-based expert reads |
Clinical Pulse
Verified clinician sentiment
Verified clinician reviews launching soon · Apply to contribute →
Reviews are signed by clinicians verified against their regulator's public register.
Inside the algorithm
Editorial featureHow Us2.ai produces a report.
Five stages — from raw input to verdict — drawn from manufacturer documentation and the public regulatory record.
- INGEST
- NORMALISE
- DETECT
- LOCALISE
- VERDICT
Stage 01 · INGEST
A whole echo study arrives as DICOM.
The software takes a complete study as DICOM objects — the 2D grayscale loops and the spectral- and tissue-Doppler traces — from a cart-based or handheld scanner. Nothing is pre-sorted: the full study is handed over as acquired, with no requirement that a sonographer label which view or measurement each clip contains.
Input
Transthoracic echocardiogram (DICOM)
Inference
Per-study, automated
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 recalls, field safety notices or FDA safety communications attributable to Us2.ai were identified in public sources as of July 2026 — read as none found rather than an exhaustive audit, since the FDA recall and MAUDE databases could not be queried directly from this environment. The clinically relevant caution is intrinsic to the task: the software measures and reports, it does not diagnose, and an automated measurement a clinician does not check is still an unchecked measurement. Handheld and novice-acquired studies carry the usual image-quality dependency, and the published accuracy figures come from studies conducted with the manufacturer's involvement and should be read with that in mind.
Intended Use & Indications
Us2.ai is image post-processing software intended to aid the diagnostic review, analysis and reporting of echocardiographic DICOM images. It automatically classifies views, segments cardiac structures and computes quantitative measurements — including left-ventricular volumes and ejection fraction, global longitudinal strain, and Doppler-derived diastolic parameters — which a qualified clinician then reviews, edits and confirms. The software supports both cart-based and handheld acquisition. Its output is decision support for the reporting clinician; it does not itself render a diagnosis, and the treating clinician remains responsible for interpretation and patient management.