Heart Failure
Conditions
Keywords
Heart failure, Artificial Intelligence-Assisted Echocardiography, Artificial Intelligence, Echocardiography, Left Ventricular Ejection Fraction
Brief summary
Heart failure (HF) is the leading cause of hospitalization among adults aged 80 years and older and represents a major diagnostic challenge in geriatric medicine due to frequently atypical clinical presentations and the presence of multiple comorbidities. Although transthoracic echocardiography (TTE) with measurement of left ventricular ejection fraction (LVEF) remains the gold standard for cardiac functional assessment, access to echocardiography is often limited in geriatric wards. Recent advances in artificial intelligence (AI) have enabled the development of portable ultrasound devices and automated image analysis software capable of providing reliable and reproducible LVEF measurements. AI-assisted automated LVEF assessment (AutoEF-AI) may therefore represent a valuable alternative to conventional echocardiography for the cardiac evaluation of older patients with heart failure.
Detailed description
Prospective, single-center interventional study comparing two methods of left ventricular ejection fraction measurement in patients aged 75 years and older hospitalized for acute heart failure.Each hemodynamically stable participant will undergo two echocardiographic examinations performed within 24 hours: * Standard Echocardiography (Reference Method) * AI-assisted automated LVEF assessment (AutoEF-AI)
Interventions
DIAGNOSTIC_TESTAutoEF-AI Assessment
Performed by a geriatrician who has completed a one-day practical training session and combines: * A handheld ultrasound device providing real-time image acquisition guidance for obtaining apical views. * Us2.ai software for automated LVEF analysis. The geriatrician is blinded to results of the gold standard echocardiography performed by the cardiologist.
DIAGNOSTIC_TESTStandard Echocardiography (Reference method)
The standard echocardiography will be performed by an expert cardiologist as part of routine clinical care and will serve as the gold standard
Sponsors
Gérond'if
Study design
Allocation
NON_RANDOMIZED
Intervention model
CROSSOVER
Primary purpose
DIAGNOSTIC
Masking
SINGLE (Investigator)
Intervention model description
Agreement between LVEF measured using AutoEF-AI and LVEF measured using standard echocardiography
Eligibility
Sex/Gender
ALL
Age
75 Years to No maximum
Healthy volunteers
No
Inclusion criteria
* Age ≥75 years. * Hospitalization in a geriatric unit for acute heart failure according to the 2021 ESC diagnostic criteria. * Hemodynamic stability at the time of echocardiographic examination. * Ability to understand study information, provide written informed consent, and willingness to participate. * Affiliation with, or beneficiary of, a health insurance/social security scheme.
Exclusion criteria
* Hemodynamic instability preventing echocardiographic assessment. * Contraindication to transthoracic echocardiography. * Patients under legal protection (guardianship, curatorship, or judicial protection measures) or unable to provide informed consent. * Refusal to participate in the study.
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Agreement between LVEF measured using AutoEF-AI and LVEF measured using standard echocardiography. | At baseline | Agreement will be assessed using: * Intraclass Correlation Coefficient (ICC) * Spearman correlation coefficient * Bland-Altman analysis * Weighted kappa coefficient for the classification of patients according to LVEF categories (≤40%, 41-49%, and ≥50%) |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Diagnostic performance of AutoEF-AI for the detection of LVEF <50%, including sensitivity, specificity, positive predictive value, negative predictive value, and overall accuracy. | At baseline | Calculation of the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall accuracy of AutoEF-AI for identifying left ventricular ejection fraction (LVEF) below 50%, compared with the standard reference method. |
| Analysis of factors associated with agreement between the two methods | At baseline | Multiple linear regression model will be used to identify clinical and technical variables (age, sex, cardiovascular history, presence of a pacemaker, cardiac arrhythmias, image quality, comorbidities, etc.) associated with a significant discrepancy between LVEF measurements obtained using AutoEF-AI and standard echocardiography. |
| Feasibility of AutoEF-AI use by a geriatrician after a short training program | At baseline | * Success rate of valid image acquisition (i.e acquizition of analyzable apical views suitable for automated analysis) * Quality of the acquired images (image quality score and proportion of non-analyzable images) * Difficulties encountered during the use of the AutoEF-AI device |
Countries
France
Contacts
CONTACTIsabelle DUFOUR
CONTACTPrisca LUCAS, PhD MPH
PRINCIPAL_INVESTIGATOROlivier HANON, MD PhD
Geriatric Department, Broca hospital
Outcome results
None listed