Disease-Specific Detection

Detect Heart Failure Across Every Phenotype

Q: What role does echocardiography play in heart failure diagnosis?

Echocardiography is the key investigation recommended by international guidelines for assessing cardiac function in heart failure. It determines the HF phenotype through LVEF measurement (HFrEF ≤40%, HFmrEF 41-49%, HFpEF ≥50%) and provides critical parameters including diastolic function, GLS, LA volume, and RV function.

Q: How does AI echocardiography help with heart failure?

AI echocardiography detects and measures every key parameter in the heart failure diagnostic pathway: LVEF, E/e', LA volume index, GLS, TAPSE, TR velocity, and LV mass index. This accurate, reproducible multi-parameter assessment distinguishes HFrEF, HFmrEF, and HFpEF, reduces measurement variability, and supports serial monitoring of treatment response.

Q: What are the different types of heart failure?

International guidelines classify heart failure into three types by left ventricular ejection fraction (LVEF): HFrEF (reduced, LVEF ≤40%), HFmrEF (mildly reduced, LVEF 41-49%), and HFpEF (preserved, LVEF ≥50%). Each type has different diagnostic criteria and treatment pathways, all determined by echocardiography.

Q: Can AI echocardiography identify the underlying cause of heart failure?

Detecting heart failure is only the first step. Us2.ai quantifies LV wall thickness, LV mass, and the regional strain pattern from the same study, and its FDA-cleared cardiac amyloidosis detection flags the apical-sparing signature of amyloid, an under-recognized cause of HFpEF. Surfacing these markers alongside the phenotype helps clinicians distinguish hypertensive heart disease, hypertrophic cardiomyopathy, and cardiac amyloidosis, conditions that each follow a different care pathway.

Q: What clinical evidence supports AI heart failure detection?

Us2.ai's automated systolic and diastolic function analysis was validated in The Lancet Digital Health (Tromp et al., 2022) against expert human readers across four independent international cohorts, detecting both reduced and preserved ejection fraction heart failure as accurately as expert sonographers with less measurement variability. Because the measurements are vendor-agnostic across major ultrasound platforms, that performance generalizes to real-world, multi-national populations rather than a single site.

Q: Are Us2.ai heart failure measurements guideline-compliant and explainable?

Yes. Us2.ai reports its measurements against ESC and ASE/EACVI recommendations, including the LVEF thresholds that define HFrEF, HFmrEF, and HFpEF. Every measurement is traceable to the contour and beat it was derived from and can be reviewed and edited, which is essential for clinical trust, audit, and regulatory accountability under FDA clearance and EU MDR.

Echocardiography defines the heart failure phenotype, HFrEF, HFmrEF, or HFpEF. Us2.ai detects and measures every guideline-recommended parameter (LVEF, GLS, diastolic function, and more) with the accuracy and reproducibility diagnosis depends on, automatically on every scan.

Book a Demo How It Works

The Clinical Problem

The World's Fastest-Growing Cardiovascular Epidemic

Heart failure affects 1–2% of adults in developed countries and over 10% of those aged 70+. Hospital admissions for HF are projected to increase 50% in the next 25 years.

64M

people affected globally

Heart failure is a global pandemic. With aging populations and increasing prevalence of risk factors, the burden is growing faster than the workforce can keep up.

~20%

1-year mortality

One in five heart failure patients die within a year of diagnosis. Five-year mortality reaches 53–67%, worse than many cancers.

>45%

readmission at 1 year

After an acute heart failure hospitalization, over 45% of patients die or are readmitted within 12 months. Early, accurate diagnosis is the first step to changing this.

Three Phenotypes, One Diagnostic Tool

International guidelines classify heart failure by LVEF. Echocardiography defines the phenotype and directly determines the treatment pathway.

≤40%

HFrEF

Reduced ejection fraction. Strongest evidence base for life-saving therapies (ARNI, beta-blockers, MRA, SGLT2i).

41–49%

HFmrEF

Mildly reduced. Emerging evidence supports HFrEF-type therapies. Requires accurate EF measurement to distinguish.

≥50%

HFpEF

Preserved. Diagnosis requires multiple echo parameters: E/e', LA volume, TR velocity, GLS, and LV mass index.

How Us2.ai Helps

Accurate Detection Across Every Heart Failure Phenotype

Heart failure detection hinges on echocardiography. Us2.ai measures every parameter in the diagnostic pathway with the accuracy and reproducibility that distinguishes HFrEF, HFmrEF, and HFpEF.

LVEF Classification

Accurate, reproducible LVEF measurement is the single most important echo parameter in heart failure. It defines the phenotype and determines the treatment pathway. Us2.ai reduces inter-observer variability.

Primary Classification

Diastolic Function Assessment

E/e' ratio, LA volume index, and TR velocity are critical for diagnosing HFpEF. Us2.ai measures all diastolic parameters on every study, essential for the most diagnostically challenging HF phenotype.

HFpEF Diagnosis

Global Longitudinal Strain

GLS detects subclinical dysfunction before LVEF drops. International guidelines note GLS <16% as a marker for HFpEF, and a ≥12% relative GLS reduction is superior to LVEF for detecting cardiotoxicity.

Early Detection

Right Heart & Structural Assessment

TAPSE, RV function, LA volume index, LV mass index, and relative wall thickness. Us2.ai delivers the full structural assessment that supports HFmrEF and HFpEF diagnosis.

Complete Workup

Serial Monitoring Built In

International guidelines recommend a follow-up 3–6 months after therapy optimization and at any clinical deterioration. Us2.ai’s automated, standardized measurements enable reliable serial comparison, detecting changes that matter and tracking treatment response over time.

Us2.ai left ventricle longitudinal measurements showing LVEF, LVEDV, and LVESV values across serial studies with trend graph

Expert Validation

Validated by Leading Heart Failure Clinicians

Us2.ai was validated against expert sonographers across four independent international cohorts, with results published in The Lancet Digital Health. Two of the field's most prominent heart failure investigators discuss what the evidence shows.

I was so skeptical. I don't think it's going to work. But I have come around to realizing that not only does it work. We absolutely need this.

Scott Solomon, MD Brigham and Women's Hospital, Harvard Medical School

If you take multiple cohorts and test against expert human readers, the message again and again is: it's as good if not better than humans. Trust the numbers.

Justin Ezekowitz, MB BCh MSc Canadian VIGOUR Centre, University of Alberta

The Lancet Digital Health 2022

Tromp J, Seekings PJ, Hung C-L, et al. Automated interpretation of systolic and diastolic function on the echocardiogram: a multicohort study

Lancet Digit Health 2022;4(1):e46–e54

A deep-learning workflow validated against expert human readers across four independent international cohorts. The AI detected reduced and preserved ejection fraction heart failure as accurately as expert sonographers, with less measurement variability, supporting reproducible, guideline-concordant assessment at scale.

Read the validation study

Landmark Publication

AI-Assisted Detection of Heart Failure from Electronic Health Records

Published with the University of Dundee, this study demonstrates how AI echocardiography can identify missed heart failure patients from electronic health records.

ESC Heart Failure 2024

Oo et al. Artificial Intelligence-Assisted Automated Heart Failure Detection and Classification from Electronic Health Records

ESC Heart Fail 2024;11:2769–2777

A collaboration with the University of Dundee demonstrating that AI-assisted echocardiography analysis can systematically identify heart failure patients who were missed through conventional clinical pathways, enabling earlier diagnosis and treatment initiation.

AI-powered HF detection: Identifies missed heart failure patients from electronic health records

Richer echo data: AI enriches dataset with critical parameters to support HF diagnosis in line with clinical guidelines

HF subtype classification: Distinguishes HFrEF from HFpEF at scale

Validated accuracy: 100% sensitivity, 94% specificity vs. manual diagnosis

Built for real-world use: Faster trial recruitment, earlier diagnosis, better surveillance

View Heart Failure Publications

Evidence & Context

Why Automated Echo Matters for Heart Failure

International guidelines place echocardiography at the center of HF diagnosis. Automation addresses the key barriers to guideline-concordant care.

Reducing LVEF Variability

International guidelines acknowledge that LVEF measurement is "subject to substantial variability." Automated analysis delivers consistent, reproducible EF, critical when treatment decisions hinge on whether EF is 39% or 41%.

Scaling with Demand

HF admissions are projected to rise 50% in 25 years. The guidelines call for more screening in asymptomatic subjects. Automated echo analysis is the only way to meet growing demand without proportionally growing the specialist workforce.

HFpEF: The Diagnostic Challenge

HFpEF remains the most difficult HF phenotype to diagnose, requiring integration of multiple echo parameters. The guidelines note "ongoing diagnostic uncertainty." Automated multi-parameter reporting helps clinicians apply the criteria consistently.

View All Heart Failure Evidence

FAQ

Frequently Asked Questions

Common questions about heart failure and AI echocardiography.

How common is heart failure?

Heart failure affects 1-2% of adults in developed countries and over 10% of those aged 70+. Approximately 64 million people are affected globally. Hospital admissions for HF are projected to increase 50% in the next 25 years due to aging populations.

What role does echocardiography play in heart failure diagnosis?

Echocardiography is the key investigation recommended by international guidelines for assessing cardiac function in heart failure. It determines the HF phenotype through LVEF measurement (HFrEF ≤40%, HFmrEF 41-49%, HFpEF ≥50%) and provides critical parameters including diastolic function, GLS, LA volume, and RV function.

How does AI echocardiography help with heart failure?

AI echocardiography detects and measures every key parameter in the heart failure diagnostic pathway: LVEF, E/e', LA volume index, GLS, TAPSE, TR velocity, and LV mass index. This accurate, reproducible multi-parameter assessment distinguishes HFrEF, HFmrEF, and HFpEF, reduces measurement variability, and supports serial monitoring of treatment response.

What are the different types of heart failure?

International guidelines classify heart failure into three types by left ventricular ejection fraction (LVEF): HFrEF (reduced, LVEF ≤40%), HFmrEF (mildly reduced, LVEF 41-49%), and HFpEF (preserved, LVEF ≥50%). Each type has different diagnostic criteria and treatment pathways, all determined by echocardiography.

Can AI echocardiography identify the underlying cause of heart failure?

Detecting heart failure is only the first step. Us2.ai quantifies LV wall thickness, LV mass, and the regional strain pattern from the same study, and its FDA-cleared cardiac amyloidosis detection flags the apical-sparing signature of amyloid, an under-recognized cause of HFpEF. Surfacing these markers alongside the phenotype helps clinicians distinguish hypertensive heart disease, hypertrophic cardiomyopathy, and cardiac amyloidosis, conditions that each follow a different care pathway.

What clinical evidence supports AI heart failure detection?

Us2.ai's automated systolic and diastolic function analysis was validated in The Lancet Digital Health (Tromp et al., 2022) against expert human readers across four independent international cohorts, detecting both reduced and preserved ejection fraction heart failure as accurately as expert sonographers with less measurement variability. Because the measurements are vendor-agnostic across major ultrasound platforms, that performance generalizes to real-world, multi-national populations rather than a single site.

Are Us2.ai heart failure measurements guideline-compliant and explainable?

Yes. Us2.ai reports its measurements against ESC and ASE/EACVI recommendations, including the LVEF thresholds that define HFrEF, HFmrEF, and HFpEF. Every measurement is traceable to the contour and beat it was derived from and can be reviewed and edited, which is essential for clinical trust, audit, and regulatory accountability under FDA clearance and EU MDR.

Echocardiography defines the HF phenotype. Accuracy makes detection reliable.

Complete Heart Failure Detection, Every Phenotype

Us2.ai measures LVEF, GLS, diastolic function, and the full structural assessment with the accuracy and reproducibility that distinguishes HFrEF, HFmrEF, and HFpEF, automatically on every scan.

Book a Demo