International evaluation of an artificial intelligence-powered electrocardiogram model detecting acute coronary occlusion myocardial infarction.

Aims: A majority of acute coronary syndromes (ACS) present without typical ST elevation. One-third of non-ST-elevation myocardial infarction (NSTEMI) patients have an acutely occluded culprit coronary artery [occlusion myocardial infarction (OMI)], leading to poor outcomes due to delayed identification and invasive management. In this study, we sought to develop a versatile artificial intelligence (AI) model detecting acute OMI on single-standard 12-lead electrocardiograms (ECGs) and compare its performance with existing state-of-the-art diagnostic criteria. Methods and results: An AI model was developed using 18 616 ECGs from 10 543 patients with suspected ACS from an international database with clinically validated outcomes. The model was evaluated in an international cohort and compared with STEMI criteria and ECG experts in detecting OMI. The primary outcome of OMI was an acutely occluded or flow-limiting culprit artery requiring emergent revascularization. In the overall test set of 3254 ECGs from 2222 patients (age 62 ± 14 years, 67% males, 21.6% OMI), the AI model achieved an area under the curve of 0.938 [95% confidence interval (CI): 0.924-0.951] in identifying the primary OMI outcome, with superior performance [accuracy 90.9% (95% CI: 89.7-92.0), sensitivity 80.6% (95% CI: 76.8-84.0), and specificity 93.7 (95% CI: 92.6-94.8)] compared with STEMI criteria [accuracy 83.6% (95% CI: 82.1-85.1), sensitivity 32.5% (95% CI: 28.4-36.6), and specificity 97.7% (95% CI: 97.0-98.3)] and with similar performance compared with ECG experts [accuracy 90.8% (95% CI: 89.5-91.9), sensitivity 73.0% (95% CI: 68.7-77.0), and specificity 95.7% (95% CI: 94.7-96.6)]. Conclusion: The present novel ECG AI model demonstrates superior accuracy to detect acute OMI when compared with STEMI criteria. This suggests its potential to improve ACS triage, ensuring appropriate and timely referral for immediate revascularization.

EDUCATE: An international, randomized controlled trial for teaching electrocardiography.

INTRODUCTION: Despite the critical role of electrocardiograms (ECGs) in patient care, evident gaps exist in ECG interpretation competency among healthcare professionals across various medical disciplines and training levels. Currently, no practical, evidence-based, and easily accessible ECG learning solution is available for healthcare professionals. The aim of this study was to assess the effectiveness of web-based, learner-directed interventions in improving ECG interpretation skills in a diverse group of healthcare professionals. METHODS: In an international, prospective, randomized controlled trial, 1206 healthcare professionals from various disciplines and training levels were enrolled. They underwent a pre-intervention test featuring 30 12-lead ECGs with common urgent and non-urgent findings. Participants were randomly assigned to four groups: (i) practice ECG interpretation question bank (question bank), (ii) lecture-based learning resource (lectures), (iii) hybrid question- and lecture-based learning resource (hybrid), or (iv) no ECG learning resources (control). After four months, a post-intervention test was administered. The primary outcome was the overall change in ECG interpretation performance, with secondary outcomes including changes in interpretation time, self-reported confidence, and accuracy for specific ECG findings. Both unadjusted and adjusted scores were used for performance assessment. RESULTS: Among 1206 participants, 863 (72 %) completed the trial. Following the intervention, the question bank, lectures, and hybrid intervention groups each exhibited significant improvements, with average unadjusted score increases of 11.4 % (95 % CI, 9.1 to 13.7; P<0.01), 9.8 % (95 % CI, 7.8 to 11.9; P<0.01), and 11.0 % (95 % CI, 9.2 to 12.9; P<0.01), respectively. In contrast, the control group demonstrated a non-significant improvement of 0.8 % (95 % CI, -1.2 to 2.8; P=0.54). While no differences were observed among intervention groups, all outperformed the control group significantly (P<0.01). Intervention groups also excelled in adjusted scores, confidence, and proficiency for specific ECG findings. CONCLUSION: Web-based, self-directed interventions markedly enhanced ECG interpretation skills across a diverse range of healthcare professionals, providing an accessible and evidence-based solution.

Comparison of diagnostic accuracy of current left bundle branch block and ventricular pacing ECG criteria for detection of occlusion myocardial infarction.

BACKGROUND: Electrocardiographic detection of patients with occlusion myocardial infarction (OMI) can be difficult in patients with left bundle branch block (LBBB) or ventricular paced rhythm (VPR) and several ECG criteria for the detection of OMI in LBBB/VPR exist. Most recently, the Barcelona criteria, which includes concordant ST deviation and discordant ST deviation in leads with low R/S amplitudes, showed superior diagnostic accuracy but has not been validated externally. We aimed to describe the diagnostic accuracy of four available ECG criteria for OMI detection in patients with LBBB/VPR at the emergency department. METHODS: The unweighted Sgarbossa criteria, the modified Sgarbossa criteria (MSC), the Barcelona criteria and the Selvester criteria were applied to chest pain patients with LBBB or VPR in a prospectively acquired database from five emergency departments. RESULTS: In total, 623 patients were included, among which 441 (71%) had LBBB and 182 (29%) had VPR. Among these, 82 (13%) patients were diagnosed with AMI, and an OMI was identified in 15 (2.4%) cases. Sensitivity/specificity of the original unweighted Sgarbossa criteria were 26.7/86.2%, for MSC 60.0/86.0%, for Barcelona criteria 53.3/82.2%, and for Selvester criteria 46.7/88.3%. In this setting with low prevalence of OMI, positive predictive values were low (Sgarbossa: 4.6%; MSC: 9.4%; Barcelona criteria: 6.9%; Selvester criteria: 9.0%) and negative predictive values were high (all >98.0%). CONCLUSIONS: Our results suggests that ECG criteria alone are insufficient in predicting presence of OMI in an ED setting with low prevalence of OMI, and the search for better rapid diagnostic instruments in this setting should continue.

A Bayesian approach to acute coronary occlusion.

In the STEMI paradigm, the disease (acute coronary occlusion) is defined and named after one element (ST elevation, without regard to the remainder of the QRST) of one imperfect test (the ECG). This leads to delayed reperfusion for patients with acute coronary occlusion whose ECGs don't meet STEMI criteria. In this editorial, we elaborate on the article by Jose Nunes de Alencar Neto about applying Bayesian reasoning to ECG interpretation. The Occlusion MI (OMI) paradigm offers evidencebased advances in ECG interpretation, expert-trained artificial intelligence, and a paradigm shift that incorporates a Bayesian approach to acute coronary occlusion.

Missing occlusions: Quality gaps for ED patients with occlusion MI.

BACKGROUND: ST-elevation Myocardial Infarction (STEMI) guidelines encourage monitoring of false positives (Code STEMI without culprit) but ignore false negatives (non-STEMI with occlusion myocardial infarction [OMI]). We evaluated the hospital course of emergency department (ED) patients with acute coronary syndrome (ACS) using STEMI vs OMI paradigms. METHODS: This retrospective chart review examined all ACS patients admitted through two academic EDs, from June 2021 to May 2022, categorized as 1) OMI (acute culprit lesion with TIMI 0-2 flow, or acute culprit lesion with TIMI 3 flow and peak troponin I >10,000 ng/L; or, if no angiogram, peak troponin >10,000 ng/L with new regional wall motion abnormality), 2) NOMI (Non-OMI, i.e. MI without OMI) or 3) MIRO (MI ruled out: no troponin elevation). Patients were stratified by admission for STEMI. Initial ECGs were reviewed for automated interpretation of "STEMI", and admission/discharge diagnoses were compared. RESULTS: Among 382 patients, there were 141 OMIs, 181 NOMIs, and 60 MIROs. Only 40.4% of OMIs were admitted as STEMI: 60.0% had "STEMI" on ECG, and median door-to-cath time was 103 min (IQR 71-149). But 59.6% of OMIs were not admitted as STEMI: 1.3% had "STEMI" on ECG (p < 0.001) and median door-to-cath time was 1712 min (IQR 1043-3960; p < 0.001). While 13.9% of STEMIs were false positive and had a different discharge diagnosis, 32.0% of Non-STEMIs had OMI but were still discharged as "Non-STEMI." CONCLUSIONS: STEMI criteria miss a majority of OMI, and discharge diagnoses highlight false positive STEMI but never false negative STEMI. The OMI paradigm reveals quality gaps and opportunities for improvement.

ECG Interpretation Proficiency of Healthcare Professionals.

ECG interpretation is essential in modern medicine, yet achieving and maintaining competency can be challenging for healthcare professionals. Quantifying proficiency gaps can inform educational interventions for addressing these challenges. Medical professionals from diverse disciplines and training levels interpreted 30 12-lead ECGs with common urgent and nonurgent findings. Average accuracy (percentage of correctly identified findings), interpretation time per ECG, and self-reported confidence (rated on a scale of 0 [not confident], 1 [somewhat confident], or 2 [confident]) were evaluated. Among the 1206 participants, there were 72 (6%) primary care physicians (PCPs), 146 (12%) cardiology fellows-in-training (FITs), 353 (29%) resident physicians, 182 (15%) medical students, 84 (7%) advanced practice providers (APPs), 120 (10%) nurses, and 249 (21%) allied health professionals (AHPs). Overall, participants achieved an average overall accuracy of 56.4% ± 17.2%, interpretation time of 142 ± 67 seconds, and confidence of 0.83 ± 0.53. Cardiology FITs demonstrated superior performance across all metrics. PCPs had a higher accuracy compared to nurses and APPs (58.1% vs 46.8% and 50.6%; P < 0.01), but a lower accuracy than resident physicians (58.1% vs 59.7%; P < 0.01). AHPs outperformed nurses and APPs in every metric and showed comparable performance to resident physicians and PCPs. Our findings highlight significant gaps in the ECG interpretation proficiency among healthcare professionals.

Education curriculum assessment for teaching electrocardiography: Rationale and design for the prospective, international, randomized controlled EDUCATE trial.

BACKGROUND: Electrocardiogram (ECG) interpretation training is a fundamental component of medical education across disciplines. However, the skill of interpreting ECGs is not universal among medical graduates, and numerous barriers and challenges exist in medical training and clinical practice. An evidence-based and widely accessible learning solution is needed. DESIGN: The EDUcation Curriculum Assessment for Teaching Electrocardiography (EDUCATE) Trial is a prospective, international, investigator-initiated, open-label, randomized controlled trial designed to determine the efficacy of self-directed and active-learning approaches of a web-based educational platform for improving ECG interpretation proficiency. Target enrollment is 1000 medical professionals from a variety of medical disciplines and training levels. Participants will complete a pre-intervention baseline survey and an ECG interpretation proficiency test. After completion, participants will be randomized into one of four groups in a 1:1:1:1 fashion: (i) an online, question-based learning resource, (ii) an online, lecture-based learning resource, (iii) an online, hybrid question- and lecture-based learning resource, or (iv) a control group with no ECG learning resources. The primary endpoint will be the change in overall ECG interpretation performance according to pre- and post-intervention tests, and it will be measured within and compared between medical professional groups. Secondary endpoints will include changes in ECG interpretation time, self-reported confidence, and interpretation accuracy for specific ECG findings. CONCLUSIONS: The EDUCATE Trial is a pioneering initiative aiming to establish a practical, widely available, evidence-based solution to enhance ECG interpretation proficiency among medical professionals. Through its innovative study design, it tackles the currently unaddressed challenges of ECG interpretation education in the modern era. The trial seeks to pinpoint performance gaps across medical professions, compare the effectiveness of different web-based ECG content delivery methods, and create initial evidence for competency-based standards. If successful, the EDUCATE Trial will represent a significant stride towards data-driven solutions for improving ECG interpretation skills in the medical community.

Machine learning for ECG diagnosis and risk stratification of occlusion myocardial infarction.

Patients with occlusion myocardial infarction (OMI) and no ST-elevation on presenting electrocardiogram (ECG) are increasing in numbers. These patients have a poor prognosis and would benefit from immediate reperfusion therapy, but, currently, there are no accurate tools to identify them during initial triage. Here we report, to our knowledge, the first observational cohort study to develop machine learning models for the ECG diagnosis of OMI. Using 7,313 consecutive patients from multiple clinical sites, we derived and externally validated an intelligent model that outperformed practicing clinicians and other widely used commercial interpretation systems, substantially boosting both precision and sensitivity. Our derived OMI risk score provided enhanced rule-in and rule-out accuracy relevant to routine care, and, when combined with the clinical judgment of trained emergency personnel, it helped correctly reclassify one in three patients with chest pain. ECG features driving our models were validated by clinical experts, providing plausible mechanistic links to myocardial injury.

Chest Pain, Paced Rhythm, and 2 Missed Indications for Emergent Reperfusion.

This case report describes a 70-year-old patient with intermittent chest pain that developed into constant chest pain, sweating, and shortness of breath.

Rapid Rule-Out of Myocardial Infarction Using a Single High-Sensitivity Cardiac Troponin I Measurement Strategy at Presentation to the Emergency Department: The SAFETY Study.

BACKGROUND: Our study addressed the diagnostic performance of the Atellica® IM High-Sensitivity Troponin I (hs-cTnI) assay for the rapid rule-out of myocardial infarction (MI) using a single hs-cTnI measurement at presentation in patients presenting to a US emergency department (ED). METHODS: This was a prospective, observational, cohort study of consecutive ED patients with suspected acute coronary syndrome, using 12-lead electrocardiogram and serial hs-cTnI measurements ordered on clinical indication (SAFETY, NCT04280926). ST-segment elevation MI patients were excluded. The optimal threshold required a sensitivity ≥99% and a negative predictive value (NPV) ≥99.5% for MI during index hospitalization as primary outcome. Type 1 MI (T1MI), myocardial injury, and 30-day adverse events were considered secondary outcomes. Event adjudications were established using the hs-cTnI assay used in clinical care. RESULTS: In 1171 patients, MI occurred in 97 patients (8.3%), 78.3% of which were type 2 MI. The optimal rule out hs-cTnI threshold was <10 ng/L, which identified 519 (44.3%) patients as low risk at presentation, with sensitivity of 99.0% (95% CI, 94.4-100) and NPV of 99.8% (95% CI, 98.9-100). For T1MI, sensitivity was 100% (95% CI, 83.9-100) and NPV 100% (95% CI, 99.3-100). Regarding myocardial injury, the sensitivity and NPV were 99.5% (95% CI, 97.9-100) and 99.8% (95% CI, 98.9-100), respectively. For 30-day adverse events, sensitivity was 96.8% (95% CI, 94.3-98.4) and NPV 97.9% (95% CI, 96.2-98.9). CONCLUSIONS: A single hs-cTnI measurement strategy enabled the rapid identification of patients at low risk of MI and 30-day adverse events, allowing potential discharge early after ED presentation. CLINICALTRIALS.GOV REGISTRATION NUMBER: NCT04280926.

Analytical performance comparing siemens whole blood point of care Atellica VTLi to the central laboratory plasma Atellica IM high-sensitivity cardiac troponin I assays.

INTRODUCTION: This study examined the analytical performance of a whole blood (WB) point of care (POC) hs-cTnI assay compared to a plasma central laboratory hs-cTnI assay in patients presenting with ischemic symptoms to a US emergency department. METHODS: Fresh WB specimens collected at 0 and 2 h from 1089 consecutive patients (2152 total from 1076 matched specimens) were analyzed for hs-cTnI using WB on POC Siemens Atellica VTLi assay and plasma on central laboratory Siemens Atellica IM assay. Concordances were determined based on concentrations ranging from < limit of detection (LoD), LoD to overall and sex specific 99th percentiles from both the IFCC manufacturer package inserts and Universal Sample Bank (USB) data, and > 99th percentiles. Method comparisons were calculated using Passing Bablok regression and Bland Altmann plots, and linear regression determined by Pearson correlation coefficient. RESULTS: Baseline concentration comparisons showed: POC VTLi < LoD 4-5 %, ≥ LoD 95 %; Atellica IM < LoD 5-7 %, and ≥ LoD 94-95 %. From the 2152 paired 0 and 2-hour samples, based on 99th percentiles, overall concordance was 91-92 % (kappa 0.72-0.77) and discordance 8 %. Passing Bablok regression analysis using 1924 specimens between LoD to 500 ng/L showed: slopes 0.469-0.490; y-intercepts 1.753-2.028; r values 0.631-0.817. Pearson correlation coefficient showed moderate to strong correlation strength, even with up to 53 % cTnI concentrations variance (Passing Bablok slopes) vs 27.0-40.1 % (Bland-Altmann plots). CONCLUSIONS: Up to 95 % of measured samples were > LoD for both the POC (Atellica VTLi) and central laboratory (Atellica IM) hs-cTnI assays. Moderate to strong concordance and correlation were observed between assays, despite up to 53 % variances in cTnI concentration.

Kenichi Harumi Plenary Address at Annual Meeting of the International Society of Computers in Electrocardiology: "What Should ECG Deep Learning Focus on? The diagnosis of acute coronary occlusion!"

According to the STEMI paradigm, only patients whose ECGs meet STEMI criteria require immediate reperfusion. This leads to reperfusion delays and significantly increases the mortality for the quarter of "non-STEMI" patients with totally occluded arteries. The Occlusion MI (OMI) paradigm has developed advanced ECG interpretation to identify this high-risk group, including examining the ECG in totality and assessing ST/T changes in proportion to the QRS. If neural networks are only developed based on STEMI databases and to identify STEMI criteria, they will simply reinforce a failed paradigm. But if deep learning is trained to identify OMI it could revolutionize patient care. This article reviews the paradigm shift from STEMI and OMI, and examines the potential and pitfalls of deep learning. This is based on the Kenichi Harumi Plenary Address at the Annual Meeting of the International Society of Computers in Electrocardiology, given by OMI expert Dr. Stephen Smith.

A two-staged classifier to reduce false positives: On device detection of atrial fibrillation using phase-based distribution of poincaré plots and deep learning.

BACKGROUND: Mobile Cardiac Outpatient Telemetry (MCOT) can be used to screen high risk patients for atrial fibrillation (AF). These devices rely primarily on algorithmic detection of AF events, which are then stored and transmitted to a clinician for review. It is critical the positive predictive value (PPV) of MCOT detected AF is high, and this often leads to reduced sensitivity, as device manufacturers try to limit false positives. OBJECTIVE: The purpose of this study was to design a two stage classifier using artificial intelligence (AI) to improve the PPV of MCOT detected atrial fibrillation episodes whilst maintaining high levels of detection sensitivity. METHODS: A low complexity, RR-interval based, AF classifier was paired with a deep convolutional neural network (DCNN) to create a two-stage classifier. The DCNN was limited in size to allow it to be embedded on MCOT devices. The DCNN was trained on 491,727 ECGs from a proprietary database and contained 128,612 parameters requiring only 158 KB of storage. The performance of the two-stage classifier was then assessed using publicly available datasets. RESULTS: The sensitivity of AF detected by the low complexity classifier was high across all datasets (>93%) however the PPV was poor (<76%). Subsequent analysis by the DCNN increased episode PPV across all datasets substantially (>11%), with only a minor loss in sensitivity (<5%). This increase in PPV was due to a decrease in the number of false positive detections. Further analysis showed that DCNN processing was only required on around half of analysis windows, offering a significant computational saving against using the DCNN as a one-stage classifier. CONCLUSION: DCNNs can be combined with existing MCOT classifiers to increase the PPV of detected AF episodes. This reduces the review burden for physicians and can be achieved with only a modest decrease in sensitivity.