ECG-AI LEF and ELEFT flag weak heart pumps with moderate accuracy

Singapore's ACE found that the Anumana and Eko AI-ECG tools were safe and moderately accurate in detecting an ejection fraction of 40% or less. Their negative predictive value was 98% to 100%.

  • Not rated a person

    Accuracy was measured against echocardiography, and in the EAGLE trial neither detection nor treatment changed to a statistically significant degree.

Depth runs from 1, a convenience or time saved, to 5, a life or a livelihood preserved.

Status
Partly reported benefit.
Evidence

foundstudiedreportedclaimed

An independent assessment report supports the benefit.

Compared with

Primary care teams without access to the AI-ECG results in the EAGLE trial. Measured.

Without the system
70/11,068 (0.6%)
With it
102/11,573 (0.9%)
The system’s part
The EAGLE trial compared primary care teams with and without access to the AI results, and the SPEC-AI Nigeria trial compared AI-guided screening with usual care. In neither trial was a gain in detection or outcomes statistically significant. ACE writes that AI-ECG uses deep learning and that both tools retrain the Mayo Clinic algorithm.
Not yet checked
LimitsCosts and harmsLearns from dataLater outcome

Download data Cite Suggest a correction

Benefit type
Improved Task Accuracy
System
Anumana ECG-AI LEF 12-Lead algorithm and Eko Low Ejection Fraction Tool (ELEFT)
Made by
Anumana, Inc. and Eko Health, Inc., both building on Mayo Clinic's 12-lead AI-ECG algorithm
Used by
Clinicians and sites in the seven studies that ACE reviewed, including the EAGLE trial's primary care teams and the SPEC-AI Nigeria trial (not in routine use in Singapore)
Where
Study sites including Nigeria, assessed for Singapore
Stage
In a pilot
Date
, when the source was published
Kind
Aggregate result
Broad benefit
Identifying Objects, Actions, and Events, Analyzing Data or Information
Collections
Scrutiny and findings, Cross-border records
Last revised
1 November 2026

Sources

  1. Assessment report, by authors with no tie to the system.

    Agency for Care Effectiveness (ACE), Ministry of Health, Singapore. Artificial intelligence-enabled electrocardiogram (AI-ECG) to detect reduced left ventricular ejection fraction. ACE Brief for New and Emerging Health Technologies, HSB-M 01/2026, dated December 2025, published on 30 April 2026. Summary of Key Points, p. 1.

    Overall, Eko and Anumana AI-ECG tools were found to be safe with moderate accuracy in detecting LVEF ≤40%.
    Findings from the EAGLE RCT showed that compared to usual care, the use of Anumana ECG-AI LEF did not result in significant changes in the detection of low LVEF (≤40%) or clinical management.

Details

AI system

Workflow

Anumana's algorithm analyzed standard 12-lead ECGs, and Eko's ELEFT analyzed single-lead ECG and heart sounds from the Eko DUO digital stethoscope. Each tool returned a binary result for an ejection fraction of 40% or less.

Follow-up

Conditions for replication

Patients are adults at risk of heart failure. Clinicians use the result with their own evaluation and clinical judgment, and an echocardiogram still confirms heart failure.

Entry details

Sector

Healthcare.

Purpose

Anumana and Eko Health built ECG-AI LEF and ELEFT to help clinicians identify adults at risk of heart failure whose left ventricular ejection fraction is 40% or less.

30 more questions not yet checked

model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.

Cite this record

Peter H. Vartanian. (2026). ECG-AI LEF and ELEFT flag weak heart pumps with moderate accuracy (AIBD-1163). Artificial Intelligence Benefits Database, edition of 1 November 2026. https://petervartanian.xyz/aibd/records/AIBD-1163.html (archived at https://doi.org/10.7910/DVN/AADSII, version 7.0)

BibTeX Data Field definitions