Prediction of Paroxysmal Atrial Fibrillation With Incorporating Genomic Information Into AI-Based ECG Analysis

Q: What are the key findings of this study?

Adding a polygenic risk score to AI-ECG did not increase AUC-ROC (0.885 vs 0.882, P=0.353) but significantly improved paroxysmal AF reclassification (NRI 0.467, IDI 0.028, P<0.001).

Q: What is the clinical evidence from this study?

Study design: Cross-Sectional. Population: Paroxysmal atrial fibrillation (n=2128). Intervention: Combined AI-based sinus rhythm ECG score and polygenic risk score (PRS) vs. AI-based sinus rhythm ECG score alone. Primary outcome: Prediction of concomitant paroxysmal atrial fibrillation (AUC-ROC and Net Reclassification Improvement) (NRI 0.467, 95% CI 0.374-0.558, p=0.353 for AUC difference; <0.001 for IDI).

Q: What does this research mean for the field?

Incorporating a polygenic risk score into AI-based ECG analysis does not significantly improve discrimination (AUC-ROC) but significantly enhances risk reclassification for paroxysmal atrial fibrillation. Novelty: ClaimNovelty.INCREMENTAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

Q: What question did this study set out to answer?

To assess whether adding polygenic risk scores and blood biomarkers improves the performance of AI-based electrocardiogram analysis for predicting paroxysmal atrial fibrillation.

Kensuke Ihara; Yuki Nagata; Kentaro Takahashi; Ken Tsuchiya; Koji Sugiyama; T Sasaki; Takatoshi Shigeta; Yasuteru Yamauchi; Naoyuki Miwa; Hitoshi Hachiya; Yasuaki Tanaka; Atsushi Takahashi; Tatsuya Hayashi; Hideo Fujita; Tomonori Watanabe; Y Imai; Masateru Takigawa; Tatsuhiko Anzai; Kunihiko Takahashi; Toshihiro Tanaka; Tetsuo Sasano; Tetsushi Furukawa

doi:10.1016/j.jacasi.2025.10.009

Key result

Adding polygenic risk score to AI-ECG improves paroxysmal AF risk reclassification despite no AUC gain.

NRI 0.467
95% CI 0.374-0.558
P=0.353 for AUC difference; <0.001 for IDI
n=2,128

Graphical abstract published by JACC Asia — Graphical abstract · Published by JACC Asia · CC BY-NC-ND 4.0

Why the study?

It remains unclear whether incorporating genetic information and blood biomarkers can improve AI models predicting paroxysmal atrial fibrillation from sinus rhythm ECGs.

Does incorporating a polygenic risk score and blood biomarkers into AI-based sinus rhythm ECG analysis improve the prediction and risk stratification of paroxysmal atrial fibrillation in Japanese cardiology patients?

Key Points

To assess whether adding polygenic risk scores and blood biomarkers improves the performance of AI-based electrocardiogram analysis for predicting paroxysmal atrial fibrillation.
Recruited 2,128 patients (1,055 with paroxysmal atrial fibrillation and 1,073 without atrial fibrillation) from 7 institutions in Japan between March 1, 2020 and December 31, 2021.
Calculated AI-based ECG scores and polygenic risk scores, and measured blood biomarkers including high-sensitivity C-reactive protein, NT-proBNP, and cell-free DNA.
Assessed predictive performance and reclassification using AUC-ROC, net reclassification improvement, and integrated discrimination improvement before and after propensity score matching.
The AI-based ECG score alone predicted paroxysmal atrial fibrillation with an AUC-ROC of 0.882 (95% CI: 0.866-0.897), whereas the polygenic risk score alone yielded an AUC-ROC of 0.655 (95% CI: 0.630-0.680).
Combining the AI-ECG score and polygenic risk score did not significantly change AUC-ROC (0.885 [95% CI: 0.870-0.901]; P = 0.353), but demonstrated significant improvements in net reclassification improvement (0.467; 95% CI: 0.374-0.558) and integrated discrimination improvement (0.028; 95% CI: 0.020-0.035; P < 0.001).
Patients in the highest quartiles of both AI-ECG score and polygenic risk score exhibited a paroxysmal atrial fibrillation prevalence of 95.8% (137 of 143), corresponding to an OR of 630.20 (95% CI: 187.90-2113.62) compared to those in the lowest quartiles for both.

Study Design

Type

Cross-Sectional (n=2,128)

Multicenter

Yes

Structured PICO

Does incorporating a polygenic risk score and blood biomarkers into AI-based sinus rhythm ECG analysis improve the prediction and risk stratification of paroxysmal atrial fibrillation in Japanese cardiology patients?

P

Population

2,128 Japanese cardiology patients over 40 years old (1,795 analyzed; median age 70 years, 32.8% female) in sinus rhythm with paroxysmal atrial fibrillation or non-AF cardiovascular conditions, evaluated cross-sectionally across 7 institutions.

E

Exposure

Combined artificial intelligence-based sinus rhythm 12-lead ECG (AI-ECG) score and polygenic risk score (PRS; PGP000392 algorithm), with evaluation of blood biomarkers (cell-free DNA [cfDNA], NT-proBNP, hs-CRP)

C

Comparator

AI-ECG score alone (and PRS alone)

O

Outcome

Prediction and risk reclassification of concomitant paroxysmal atrial fibrillation (PAF), assessed by area under the receiver-operating characteristic curve (AUC-ROC), continuous net reclassification improvement (NRI), and integrated discrimination improvement (IDI)surrogate

Main Result

Effect estimate: NRI 0.467 (95% CI 0.374-0.558)

Absolute Event Rate: 0.885% vs 0.882%

p-value: p=0.353 for AUC difference; <0.001 for IDI

Incorporating genomic polygenic risk scores (and cell-free DNA) into sinus-rhythm AI-ECG analysis significantly improves risk reclassification and discrimination for identifying patients with paroxysmal atrial fibrillation despite not increasing overall AUC-ROC.

Limitations

Relatively small sample size requiring validation in larger cohorts
AI models and PRS algorithms were developed/validated using Japanese data, limiting generalizability outside Japan
Lack of longitudinal follow-up to assess future AF development
Clinical background data other than blood biomarkers were not incorporated into predictive models due to selection bias
Overlap of ECGs with the original AI model development cohort may have overestimated AI-ECG performance
AI models and PRS algorithms were developed or validated using Japanese data, limiting applicability outside Japan
No longitudinal follow-up to assess future development of AF
Clinical background data other than blood biomarkers were not incorporated into the predictive models due to selection bias
Overlap of study ECGs with the original AI model development cohort may have overestimated AI-ECG performance and underestimated the added value of PRS

Ask about this paper

Answers are grounded in this paper and cite their sources.

Implication

Should not yet change AI-ECG use for paroxysmal AF; leaves open whether reclassification gains warrant prospective validation.

Abstract

BACKGROUND: It remains unclear whether incorporating genetic information and blood biomarkers can improve artificial intelligence (AI) models predicting paroxysmal atrial fibrillation (PAF) from sinus rhythm electrocardiograms (ECGs). OBJECTIVES: This study aimed to explore the potential value of polygenic risk scores (PRS) and blood biomarkers in AI-based ECG (AI-ECG) analysis for predicting PAF. METHODS: Patients from 7 institutions in Japan were recruited between March 1, 2020 and December 31, 2021. AI-ECG scores and PRS were calculated using previously reported AI model and PRS algorithm. High-sensitivity C-reactive protein, N-terminal pro-B-type natriuretic peptide, and cell-free DNA (cfDNA) levels were also evaluated. RESULTS: A total of 2,128 cases were included (PAF: 1,055 of 2,128 [49.6%]). AI-ECG scores were significantly higher in PAF group than in non-AF group, with an area under the curve (AUC) of 0.882 (95% CI: 0.866-0.897) for the receiver-operating characteristic (ROC) curve in predicting PAF. PRS was also significantly higher in PAF group, with AUC-ROC of 0.655 (95% CI: 0.630-0.680). Although combining AI-ECG scores and PRS did not improve AUC-ROC, a significant improvement was observed in the net reclassification improvement of 0.467 (95% CI: 0.374-0.558) and integrated discrimination improvement of 0.028 (95% CI: 0.020-0.035). Among blood biomarkers, only cell-free DNA, along with PRS and AI-ECG, was associated with PAF in multivariable analysis; however, its contribution to PAF prediction was limited. CONCLUSIONS: Genetic information may provide complementary insights and improve risk stratification for the prediction of PAF using AI-ECG.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Ihara et al. (2026) conducted a cross-sectional in Paroxysmal atrial fibrillation (n=2,128). Combined AI-based sinus rhythm ECG score and polygenic risk score (PRS) vs. AI-based sinus rhythm ECG score alone was evaluated on Prediction of concomitant paroxysmal atrial fibrillation (AUC-ROC and Net Reclassification Improvement) (NRI 0.467, 95% CI 0.374-0.558, p=0.353 for AUC difference; <0.001 for IDI). Adding a polygenic risk score to AI-ECG did not increase AUC-ROC (0.885 vs 0.882, P=0.353) but significantly improved paroxysmal AF reclassification (NRI 0.467, IDI 0.028, P<0.001).

synapsesocial.com/papers/69d9e47378050d08c1b74fdb https://doi.org/10.1016/j.jacasi.2025.10.009

View Full Paper

Ask AI

Bookmark

Share

Prediction of Paroxysmal Atrial Fibrillation With Incorporating Genomic Information Into AI-Based ECG Analysis

Key Points

Study Design

Structured PICO

Main Result

Limitations

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

Prediction of Paroxysmal Atrial Fibrillation With Incorporating Genomic Information Into AI-Based ECG Analysis

Key Points

Study Design

Structured PICO

Main Result

Limitations

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

Keep reading

Keep reading