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.

Thermodilution-derived volumetric resting coronary blood flow measurement in humans.

BACKGROUND: Quantification of microvascular function requires the measurement of flow and resistance at rest and during hyperaemia. Continuous intracoronary thermodilution accurately measures coronary flow during hyperaemia. AIMS: The aim of this study was to investigate whether continuous coronary thermodilution using lower infusion rates also enables volumetric coronary blood flow measurements (in mL/min) at rest. METHODS: In 59 patients (88 arteries), the ratio of distal to proximal coronary pressure (Pd/Pa), as well as absolute blood flow (in mL/min) by continuous thermodilution, was recorded using a pressure/temperature guidewire. Saline was infused at rates of 10 and 20 mL/min. In 27 arteries, Doppler average peak velocity (APV) was measured simultaneously. Pd/Pa, APV, thermodilution-derived coronary flow reserve (CFRthermo) and coronary flow velocity reserve (CFVR) were assessed. In 10 arteries, simultaneous recordings were obtained at saline infusion rates of 6, 8, 10 and 20 mL/min. RESULTS: Compared to baseline, saline infusion at 10 mL/min did not change Pd/Pa (0.95±0.05 versus 0.94±0.05, p=0.49) or APV (22±8 versus 23±8 cm/s, p=0.60); conversely, an infusion rate of 20 mL/min induced a decrease in Pd/Pa and an increase in APV. Stable thermodilution tracings were obtained during saline infusion at 8 and 10 mL/min, but not at 6 mL/min. Mean values of CFRthermo and CFVR were similar (2.78±0.91 versus 2.76±1.06, p=0.935) and their individual values correlated closely (r=0.89, 95% CI: 0.78-0.95, p<0.001). CONCLUSIONS: In addition to hyperaemic flow, continuous thermodilution can quantify absolute resting coronary blood flow; therefore, it can be used to calculate coronary flow reserve and microvascular resistance reserve.

Evolving concepts of angiogram: fractional flow reserve discordances in 4000 coronary stenoses.

AIMS: The present analysis addresses the potential clinical and physiologic significance of discordance in severity of coronary artery disease between the angiogram and fractional flow reserve (FFR) in a large and unselected patient population. METHODS AND RESULTS: Between September 1999 and December 2011, FFR and percent diameter stenosis (DS) as assessed by quantitative coronary angiography were obtained in 2986 patients (n = 4086 coronary stenoses), in whom at least one stenosis was of intermediate angiographic severity. Fractional flow reserve correlated slightly but significantly with DS [-0.38 (95% CI: -0.41; -0.36); P < 0.001]. The sensitivity, specificity, and diagnostic accuracy of a ≥ 50% DS for predicting FFR ≤ 0.80 were 61% (95% CI: 59; 63), 67% (95% CI: 65; 69), and 0.64 (95% CI: 0.56; 0.72), respectively. In different anatomical settings, sensitivity and specificity showed marked variations between 35 to 74% and 58 to 76%, respectively, resulting in a discordance in 35% of all cases for these thresholds. For an angiographic threshold of 70% DS, the diagnostic performance by the Youden's index decreased from 0.28 to 0.11 for the overall population. CONCLUSION: The data confirm that one-third of a large patient population shows discordance between angiogram ≥ 50%DS and FFR ≤ 0.8 thresholds of stenosis severity. Left main stenoses are often underestimated by the classical 50% DS cut-off compared with FFR. This discordance offers physiologic insights for future trials. It is hypothesized that the discordance between angiography and FFR is related to technical limitations, such as imprecise luminal border detection by angiography, as well as to physiologic factors, such as variable minimal microvascular resistance.

Long-term clinical outcome after fractional flow reserve-guided treatment in patients with angiographically equivocal left main coronary artery stenosis.

BACKGROUND: Significant left main coronary artery stenosis is an accepted indication for surgical revascularization. The potential of angiography to evaluate the hemodynamic severity of a stenosis is limited. The aims of the present study were to assess the long-term clinical outcome of patients with an angiographically equivocal left main coronary artery stenosis in whom the revascularization strategy was based on fractional flow reserve (FFR) and to determine the relationship between quantitative coronary angiography and FFR. METHODS AND RESULTS: In 213 patients with an angiographically equivocal left main coronary artery stenosis, FFR measurements and quantitative coronary angiography were performed. When FFR was > or =0.80, patients were treated medically or another stenosis was treated by coronary angioplasty (nonsurgical group; n=138). When FFR was <0.80, coronary artery bypass grafting was performed (surgical group; n=75). The 5-year survival estimates were 89.8% in the nonsurgical group and 85.4% in the surgical group (P=0.48). The 5-year event-free survival estimates were 74.2% and 82.8% in the nonsurgical and surgical groups, respectively (P=0.50). Percent diameter stenosis at quantitative coronary angiography correlated significantly with FFR (r=-0.38, P<0.001), but a very large scatter was observed. In 23% of patients with a diameter stenosis <50%, the left main coronary artery stenosis was hemodynamically significant by FFR. CONCLUSIONS: In patients with equivocal stenosis of the left main coronary artery, angiography alone does not allow appropriate individual decision making about the need for revascularization and often underestimates the functional significance of the stenosis. The favorable outcome of an FFR-guided strategy suggests that FFR should be assessed in such patients before a decision is made "blindly" about the need for revascularization.