PulseTrendingJournal ClubResearchersJournalsExplore
Instagram
HomeTrendingJournal ClubExplore
Synapse
⌘+K
Synapse
July 11, 2025Open Access

Artificial Intelligence-Enhanced Electrocardiogram Models for Detection of Left Ventricular Dysfunction: A Comparison Study

View Full Paper
Ask AI
Bookmark
Share

Why the study?

Head-to-head agreement and performance of AI-ECG models for detecting left ventricular systolic dysfunction had not been independently compared within the same cohort.

Do artificial intelligence-enhanced electrocardiogram models accurately detect left ventricular systolic dysfunction in patients undergoing routine CMR?

Population

1,203 consecutive patients undergoing routine clinical CMR with paired ECGs

Comparison

Head-to-head comparison of four shared AI-ECG models

Design

Systematic review and independent external validation study

Authors

PCPhilip M. CroonMBMachteld BoonstraCACornelis P. Allaart

Discussion

Loading...

Member takes

Overview

Variable AI-ECG performance for LVSD warrants clinical caution; leaves open reproducibility standards and model selection pending validation.

Structured PICO

Do artificial intelligence-enhanced electrocardiogram models accurately detect left ventricular systolic dysfunction in patients undergoing routine CMR?

P
Population
1,203 patients undergoing routine clinical cardiac magnetic resonance imaging (CMR) with paired 12-lead ECGs within a 45-day window, mean age 59 ± 15 years, 36% female.
I
Intervention
Four artificial intelligence-enhanced electrocardiogram (AI-ECG) models (AiTiALVSD, ECG Vision, CGMH, Utrecht) for detecting left ventricular systolic dysfunction.
C
Comparator
Head-to-head comparison of the models, using CMR-derived LVEF as the reference standard.
O
Outcome
Area under the receiver operating characteristic curve (AUROC) for detecting left ventricular systolic dysfunction.surrogate

AI-ECG models demonstrate strong and consistent performance for detecting left ventricular systolic dysfunction across disparate populations, though lack of model sharing limits independent validation.

Limitations

  • Limited availability of models hinders independent validation

Cite This Study

Croon et al. (2025) studied this question.

synapsesocial.com/papers/6a7cc7c603b6c06a674c7f0fhttps://doi.org/10.1101/2025.07.08.25331129
View Full Paper
Ask AI
Bookmark
Share