Reanalyze Your Echo Archive
Run studies already in your archive through Us2.ai's automated measurements and guideline-based report conclusions, so a care program can find patients whose echo met a guideline threshold and who were never followed up. A clinician reviews every result.
Years of Echos Already in Your Archive
Every health system holds years of archived echocardiograms in PACS. Reprocessed automatically, they give a care program a list of patients to review.
- Heart failure identification and classification from health records with the University of Dundee, using Us2.ai's automated echo analysis (ESC Heart Failure 2024)
- Research: archived echos from over 11,000 consecutive patients screened for cardiac amyloidosis (AI-SCREEN-CA, AHA 2025), with 82.8% sensitivity and 95.2% specificity
- In that study, amyloidosis prevalence was over five times higher in screen-positive than in screen-negative patients (16.7% vs. 3.0%)
- No manual step per study, so the size of the archive is not limited by reader time
- Results from the archive stay with the patient record and are shown when the patient returns for a new echo
Echo software that needs a person to select views, adjust tracings or confirm measurements on each study cannot reasonably run over thousands of archived studies. Us2.ai produces the measurements and the draft report with no manual step, and a clinician reviews only the studies the program selects.
From Archive to a Review List
Us2.ai connects to your PACS, CVIS or EMR, reprocesses archived echocardiograms through the same automated analysis as a new study, and lists the studies that meet the criteria your program sets.
Connect to Your Archive
Us2.ai integrates with PACS, CVIS, EMR or echo cart storage, deployed in the cloud or on-premise.
Automated Reanalysis
Each archived study goes through view classification, measurement, guideline-based report conclusions and a structured report, with no manual step per study.
Studies for Clinician Review
Your program sets the criteria, for example a reduced ejection fraction, a severe aortic stenosis grade or an Us2.ca flag in an eligible patient. Matching studies go to a clinician with the measurements behind each result.
Serial Measurements
Every study gets the same structured report, so a patient's studies can be compared over time and archived results connected to the next visit.
When a patient selected from the archive returns for a new echocardiogram, the clinician sees the earlier measurements and report conclusions beside the new ones, so follow-up starts from the earlier study.
What Research Has Shown
These studies used Us2.ai's automated analysis on archived echocardiograms. Screening is a research use, not a cleared indication.
Artificial Intelligence-Assisted Automated Heart Failure Detection and Classification from Electronic Health Records
With the University of Dundee, this study used Us2.ai's automated echocardiographic analysis to identify and classify heart failure from electronic health records.
AI-SCREEN-CA: Amyloidosis Screening at Scale
A late-breaking research presentation, not yet a peer-reviewed paper. AI-based cardiac amyloidosis screening of archived echos from over 11,000 consecutive patients reached 82.8% sensitivity and 95.2% specificity, with prevalence over five times higher in screen-positive than screen-negative patients (16.7% vs. 3.0%).
Automated Measurement Against Expert Readers
In a validation set of 602 studies, each measured three times by core lab sonographers, the disagreement between Us2.ai and the sonographers was lower than the disagreement between sonographers, for every parameter tested.
Frequently Asked Questions
Common questions about retrospective EHR/PACS mining with Us2.ai.
What is EHR mining with echocardiography AI?
Reprocessing echocardiograms already in your archive (PACS, CVIS or EMR) through Us2.ai. Each archived study gets the same automated measurements and guideline-based report conclusions as a new one, for a clinician to review. A care program can then find patients whose archived echo met a guideline threshold but who have no matching diagnosis or follow-up.
Why does automation matter for retrospective echo analysis?
Reprocessing thousands of archived studies is practical only when nobody has to select views or trace each one. Us2.ai runs view classification, measurement and the draft report with no manual step per study, and a clinician reviews the studies the program selects.
What can be found in archived echos with Us2.ai?
Measurements and guideline-based report conclusions: heart failure category, aortic stenosis grade and the echocardiographic probability of pulmonary hypertension. Where Us2.ca is labeled, it flags studies suggestive of cardiac amyloidosis in adults with an IVSd or LVPWd of 12 mm or more. None of these is a diagnosis, and each is for clinician review.
Has retrospective echo analysis with Us2.ai been studied?
Yes, in research. Oo et al. (ESC Heart Failure 2024) used Us2.ai's automated echo analysis to identify and classify heart failure from health records with the University of Dundee. AI-SCREEN-CA (Sakamoto, Kagiyama et al., late-breaking presentation, AHA Scientific Sessions 2025) screened archived echos from more than 11,000 consecutive patients for cardiac amyloidosis, with 82.8% sensitivity and 95.2% specificity. Both are research uses, not cleared indications. The labeled Us2.ca performance is on the cardiac amyloidosis page.
Archived echos can be reanalyzed with the same automated analysis as new ones.
Put Your Echo Archives to Work
Reprocess archived studies automatically and give your care program a list of patients for clinician review.