Validation of an automated artificial intelligence system for 12­lead ECG interpretation.

BACKGROUND: The electrocardiogram (ECG) is one of the most accessible and comprehensive diagnostic tools used to assess cardiac patients at the first point of contact. Despite advances in computerized interpretation of the electrocardiogram (CIE), its accuracy remains inferior to physicians. This study evaluated the diagnostic performance of an artificial intelligence (AI)-powered ECG system and compared its performance to current state-of-the-art CIE. METHODS: An AI-powered system consisting of 6 deep neural networks (DNN) was trained on standard 12­lead ECGs to detect 20 essential diagnostic patterns (grouped into 6 categories: rhythm, acute coronary syndrome (ACS), conduction abnormalities, ectopy, chamber enlargement and axis). An independent test set of ECGs with diagnostic consensus of two expert cardiologists was used as a reference standard. AI system performance was compared to current state-of-the-art CIE. The key metrics used to compare performances were sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and F1 score. RESULTS: A total of 932,711 standard 12­lead ECGs from 173,949 patients were used for AI system development. The independent test set pooled 11,932 annotated ECG labels. In all 6 diagnostic categories, the DNNs achieved high F1 scores: Rhythm 0.957, ACS 0.925, Conduction abnormalities 0.893, Ectopy 0.966, Chamber enlargement 0.972, and Axis 0.897. The diagnostic performance of DNNs surpassed state-of-the-art CIE for the 13 out of 20 essential diagnostic patterns and was non-inferior for the remaining individual diagnoses. CONCLUSIONS: Our results demonstrate the AI-powered ECG model's ability to accurately identify electrocardiographic abnormalities from the 12­lead ECG, highlighting its potential as a clinical tool for healthcare professionals.

Utilizing longitudinal data in assessing all-cause mortality in patients hospitalized with heart failure.

AIMS: Risk stratification in patients with a new onset or worsened heart failure (HF) is essential for clinical decision making. We have utilized a novel approach to enrich patient level prognostication using longitudinally gathered data to develop ML-based algorithms predicting all-cause 30, 90, 180, 360, and 720 day mortality. METHODS AND RESULTS: In a cohort of 2449 HF patients hospitalized between 1 January 2011 and 31 December 2017, we utilized 422 parameters derived from 151 451 patient exams. They included clinical phenotyping, ECG, laboratory, echocardiography, catheterization data or percutaneous and surgical interventions reflecting the standard of care as captured in individual electronic records. The development of predictive models consisted of 101 iterations of repeated random subsampling splits into balanced training and validation sets. ML models yielded area under the receiver operating characteristic curve (AUC-ROC) performance ranging from 0.83 to 0.89 on the outcome-balanced validation set in predicting all-cause mortality at aforementioned time-limits. The 1 year mortality prediction model recorded an AUC of 0.85. We observed stable model performance across all HF phenotypes: HFpEF 0.83 AUC, HFmrEF 0.85 AUC, and HFrEF 0.86 AUC, respectively. Model performance improved when utilizing data from more hospital contacts compared with only data collected at baseline. CONCLUSIONS: Our findings present a novel, patient-level, comprehensive ML-based algorithm for predicting all-cause mortality in new or worsened heart failure. Its robust performance across phenotypes throughout the longitudinal patient follow-up suggests its potential in point-of-care clinical risk stratification.