What is the clinical evidence from this study?

Study design: Cohort. Population: Atrial fibrillation (n=30467). Intervention: Artificial Intelligence (AI) model for 12-lead ECG. Primary outcome: Prediction of atrial fibrillation onset within 1 year from sinus rhythm ECG (Pre-AF) (AUC 0.846).

What question did this study set out to answer?

To assess the efficacy of an artificial intelligence model in detecting atrial fibrillation during sinus rhythm using ECG readings.

May 12, 2026Health Evaluation and Promotion0 citationsOpen Access

Efficacy of Artificial Intelligence on 12-leads ECG to Detect Atrial Fibrillation during Sinus Rhythm

MTMasaki TakeuchiFujitsu (China)YMYoshimasa MISHUKUFujitsu (China)EHEriko HasumiElectrophysiology

Key Result

An AI model predicting atrial fibrillation onset within 1 year from 12-lead ECGs during sinus rhythm achieved an AUC of 0.846 with 44.0% sensitivity and 95.7% specificity in a Japanese cohort.

Key Points

To assess the efficacy of an artificial intelligence model in detecting atrial fibrillation during sinus rhythm using ECG readings.
Developed AI for screening atrial fibrillation from non-episodic 12-lead ECGs
Target population included Japanese individuals
Clinical evaluation of AI effectiveness in a real-world setting
AI demonstrated enhanced accuracy in detecting atrial fibrillation compared to traditional methods.
Significant reduction in false negatives observed with the use of AI (exact metrics not specified).

Study Design

Type

Cohort (n=30,467)

Multicenter

Structured PICO

Does an artificial intelligence model applied to 12-lead ECG during sinus rhythm detect paroxysmal atrial fibrillation in Japanese individuals?

Population

Japanese individuals

Intervention

Artificial intelligence (AI) applied to 12-lead ECG during sinus rhythm

Outcome

Detection of paroxysmal atrial fibrillation

This study aims to develop and validate an AI model for detecting paroxysmal atrial fibrillation from 12-lead ECGs during sinus rhythm in a Japanese population.

Main Result

Effect estimate: AUC 0.846

Limitations

Evaluation was based on retrospective data from a single facility, requiring future validation with data from other institutions.
Need for prospective data evaluation to confirm accuracy.
High number of false positives when applied to the general population, necessitating combination with other tests to reduce false positive rates.

Abstract

【目的】心房細動（atrial fibrillation，AF）は脳梗塞、心不全や死亡のリスクを増加させることから、早期の診断と抗凝固療法の導入が必要となるが、AFは無症状のことも多く、早期発見が難しい。そこで、本研究は日本人を対象とし、AFの早期発見のため非発作時の心電図から発作性AFをスクリーニングするための人工知能（artificial intelligence，AI）開発を行い、本AIの臨床の場における有効性について検証することを目的とした。

AIに質問

Bookmark

View Full Paper

AIに質問

Bookmark

View Full Paper

Cite This Study

竹内ら（火曜日）は心房細動（n=30,467）のコホートを実施しました。12リードECG用の人工知能（AI）モデルは、洞調律ECGから1年以内の心房細動発症の予測（Pre-AF）について評価されました（AUC 0.846）。洞調律中の12リードECGから1年以内の心房細動発症を予測するAIモデルは、日本のコホートにおいて、感度44.0%、特異度95.7%でAUC 0.846を達成しました。

synapsesocial.com/papers/6a02c2fdce8c8c81e9640459 https://doi.org/https://doi.org/10.7143/jhep.53.330

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: