Artificial Intelligence–Enabled Prediction of Heart Failure Risk From Single-Lead Electrocardiograms DOI
Lovedeep S Dhingra, Arya Aminorroaya, Aline F Pedroso

et al.

JAMA Cardiology, Journal Year: 2025, Volume and Issue: unknown

Published: April 16, 2025

Importance Despite the availability of disease-modifying therapies, scalable strategies for heart failure (HF) risk stratification remain elusive. Portable devices capable recording single-lead electrocardiograms (ECGs) may enable large-scale community-based assessment. Objective To evaluate whether an artificial intelligence (AI) algorithm can predict HF from noisy ECGs. Design, Setting, and Participants A retrospective cohort study individuals without at baseline was conducted among with conventionally obtained outpatient ECGs in integrated Yale New Haven Health System (YNHHS) prospective population-based cohorts UK Biobank (UKB) Brazilian Longitudinal Study Adult (ELSA-Brasil). Data analysis performed September 2023 to February 2025. Exposure AI-ECG–defined left ventricular systolic dysfunction (LVSD). Main Outcomes Measures Among ECGs, lead I were isolated a noise-adapted AI-ECG model (to simulate ECG signals wearable devices) trained identify LVSD deployed. The association probability new-onset HF, defined as first hospitalization, evaluated. discrimination compared against 2 scores (Pooled Cohort Equations Prevent Heart Failure [PCP-HF] Predicting Risk Cardiovascular Disease Events [PREVENT] equations) using Harrel C statistic, improvement, net reclassification improvement. Results There 192 667 YNHHS patients (median [IQR] age, 56 [41-69] years; 111 181 women [57.7%]), 42 141 UKB participants 65 [59-71] 21 795 [51.7%]), 13 454 ELSA-Brasil 51 [45-58] 7348 [54.6%]) total 3697 (1.9%) developed over median (IQR) 4.6 (2.8-6.6) years, 46 (0.1%) 3.1 (2.1-4.5) 31 (0.2%) 4.2 (3.7-4.5) years. positive screening result associated 3- 7-fold higher each 0.1 increment 27% 65% hazard across cohorts, independent sex, comorbidities, competing death. AI-ECG’s 0.723 (95% CI, 0.694-0.752) YNHHS, 0.736 0.606-0.867) UKB, 0.828 0.692-0.964) ELSA-Brasil. Across incorporating predictions alongside PCP-HF PREVENT equations statistic (difference addition PCP-HF, 0.080-0.107; difference PREVENT, 0.069-0.094). had improvement 0.091 0.205 vs 0.068 0.192 PREVENT; it 18.2% 47.2% 11.8% 47.5% PREVENT. Conclusions Relevance multinational estimated suggesting potential risk-stratification strategy requiring portable devices.

Language: Английский

Harnessing Artificial Intelligence for Innovation in Interventional Cardiovascular Care DOI Creative Commons
Arya Aminorroaya, Dhruva Biswas, Aline F Pedroso

et al.

Journal of the Society for Cardiovascular Angiography & Interventions, Journal Year: 2025, Volume and Issue: 4(3), P. 102562 - 102562

Published: March 1, 2025

Artificial intelligence (AI) serves as a powerful tool that can revolutionize how personalized, patient-focused care is provided within interventional cardiology. Specifically, AI augment clinical across the spectrum for acute coronary syndrome, artery disease, and valvular heart with applications in structural interventions. This has been enabled by potential of to harness various types health data. We review AI-driven technologies advance diagnosis, preprocedural planning, intraprocedural guidance, prognostication automates tasks, increases efficiency, improves reliability accuracy, individualizes care, establishing its transform care. Furthermore, AI-enabled, community-based screening programs are yet be implemented leverage full improve patient outcomes. However, practice, tools require robust transparent development processes, consistent performance settings populations, positive impact on quality outcomes, seamless integration into workflows. Once these established, reshape cardiology, improving precision,

Language: Английский

Citations

1
Implementing ECG-AI in Clinical Care DOI
Minjae Yoon, Seng Chan You

Journal of the American College of Cardiology, Journal Year: 2025, Volume and Issue: 85(12), P. 1314 - 1316

Published: March 24, 2025

Language: Английский

Citations

0
Artificial Intelligence–Enabled Prediction of Heart Failure Risk From Single-Lead Electrocardiograms DOI
Lovedeep S Dhingra, Arya Aminorroaya, Aline F Pedroso

et al.

JAMA Cardiology, Journal Year: 2025, Volume and Issue: unknown

Published: April 16, 2025

Importance Despite the availability of disease-modifying therapies, scalable strategies for heart failure (HF) risk stratification remain elusive. Portable devices capable recording single-lead electrocardiograms (ECGs) may enable large-scale community-based assessment. Objective To evaluate whether an artificial intelligence (AI) algorithm can predict HF from noisy ECGs. Design, Setting, and Participants A retrospective cohort study individuals without at baseline was conducted among with conventionally obtained outpatient ECGs in integrated Yale New Haven Health System (YNHHS) prospective population-based cohorts UK Biobank (UKB) Brazilian Longitudinal Study Adult (ELSA-Brasil). Data analysis performed September 2023 to February 2025. Exposure AI-ECG–defined left ventricular systolic dysfunction (LVSD). Main Outcomes Measures Among ECGs, lead I were isolated a noise-adapted AI-ECG model (to simulate ECG signals wearable devices) trained identify LVSD deployed. The association probability new-onset HF, defined as first hospitalization, evaluated. discrimination compared against 2 scores (Pooled Cohort Equations Prevent Heart Failure [PCP-HF] Predicting Risk Cardiovascular Disease Events [PREVENT] equations) using Harrel C statistic, improvement, net reclassification improvement. Results There 192 667 YNHHS patients (median [IQR] age, 56 [41-69] years; 111 181 women [57.7%]), 42 141 UKB participants 65 [59-71] 21 795 [51.7%]), 13 454 ELSA-Brasil 51 [45-58] 7348 [54.6%]) total 3697 (1.9%) developed over median (IQR) 4.6 (2.8-6.6) years, 46 (0.1%) 3.1 (2.1-4.5) 31 (0.2%) 4.2 (3.7-4.5) years. positive screening result associated 3- 7-fold higher each 0.1 increment 27% 65% hazard across cohorts, independent sex, comorbidities, competing death. AI-ECG’s 0.723 (95% CI, 0.694-0.752) YNHHS, 0.736 0.606-0.867) UKB, 0.828 0.692-0.964) ELSA-Brasil. Across incorporating predictions alongside PCP-HF PREVENT equations statistic (difference addition PCP-HF, 0.080-0.107; difference PREVENT, 0.069-0.094). had improvement 0.091 0.205 vs 0.068 0.192 PREVENT; it 18.2% 47.2% 11.8% 47.5% PREVENT. Conclusions Relevance multinational estimated suggesting potential risk-stratification strategy requiring portable devices.

Language: Английский

Citations

0