Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: May 24, 2026

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

Interpreting ECG Images with Multimodal Large Language Models.

Keyuan Jiang¹, Japp Adhikari¹, Gordon R Bernard²

¹Purdue University Northwest, Hammond, Indiana, U.S.A.

Studies in Health Technology and Informatics

|May 23, 2026

Summary

This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Conducting multicenter trials through the trial innovation network comprehensive consultation.

Contemporary clinical trials·2026

Same author

Toward Personalized Rotator Cuff Physical Therapy Dosage Using a Machine Learning-Based Pilot Study with EMG.

Bioengineering (Basel, Switzerland)·2026

Same author

Design of the BEST-II Randomized Clinical Trial.

Stroke (Hoboken, N.J.)·2026

Same author

A REDCap advanced randomization module to meet the needs of modern trials.

Journal of biomedical informatics·2025

Same author

Statins for polycystic ovary syndrome in varying resource settings: a phenome-wide association study and evidence synthesis.

Frontiers in pharmacology·2025

Same author

A Neural Embedding Approach to Mapping Health Concepts to Concept Unique Identifiers.

Studies in health technology and informatics·2025

Same journal

A GenAI Pipeline for Violinist Kinematic Data Management.

Studies in health technology and informatics·2026

Same journal

AMAL-For-Qatar: A Comprehensive AI Ecosystem for Fetal Ultrasound Analysis - Project Overview and Achievements.

Studies in health technology and informatics·2026

Same journal

Longitudinal Treatment-Aware Multimodal AI for Dermatology: A Scoping Review.

Studies in health technology and informatics·2026

Same journal

Predicting Postpartum Depression Using Imbalance-Aware Machine Learning.

Studies in health technology and informatics·2026

Same journal

Validation of Deep-Learning Models for Autosegmentation of Brain Metastases.

Studies in health technology and informatics·2026

Same journal

Delay-Dependent Gating in Modular RNNs.

Studies in health technology and informatics·2026

See all related articles

Large language models show promise in interpreting electrocardiography (ECG) images for cardiac conditions. AI models like GPT-4o and Gemini 2.5 Pro achieved notable accuracy in few-shot learning tasks for disease detection.

Area of Science:

Cardiology
Artificial Intelligence
Medical Imaging

Background:

Cardiac diseases are a leading global cause of mortality.
Electrocardiography (ECG) is crucial for diagnosing conditions like myocardial infarction and arrhythmias.
Advancements in AI, particularly multimodal large language models (LLMs), offer potential for enhanced ECG interpretation.

Purpose of the Study:

To evaluate the diagnostic accuracy of Google Gemini 2.5 Pro and OpenAI GPT-4o in interpreting ECG images.
To assess the effectiveness of few-shot learning (5-shot and 10-shot classification) for cardiac condition recognition using LLMs.

Main Methods:

A corpus of 928 annotated ECG images was used for evaluation.
Two few-shot learning approaches were employed: 5-shot (examples aligned with medical resident annotations) and 10-shot (randomly selected training examples).

Keywords:

Cardiac disease diagnosis ECG images few-shot classification image understanding multimodal large language model

Related Experiment Videos

Last Updated: May 24, 2026

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

The performance of Gemini 2.5 Pro and GPT-4o was compared based on classification accuracy.

Main Results:

Few-shot learning significantly improved ECG interpretation accuracy for both models.
GPT-4o achieved higher accuracies: 0.819 (10-shot) and 0.829 (5-shot).
Gemini 2.5 Pro achieved accuracies of 0.635 (10-shot) and 0.543 (5-shot).

Conclusions:

Multimodal large language models demonstrate significant potential for accurate ECG image interpretation.
These AI models may offer a cost-effective solution for diagnosing cardiac conditions from ECGs.
Further research into LLM applications in medical diagnostics is warranted.