Physician accuracy in interpreting potential ST-segment elevation myocardial infarction electrocardiograms

Artificial neural networks for ECG interpretation in acute coronary syndrome: A scoping review

INTRODUCTION

The electrocardiogram (ECG) is a crucial diagnostic tool in the Emergency Department (ED) for assessing patients with Acute Coronary Syndrome (ACS). Despite its widespread use, the ECG has limitations, including low sensitivity of the STEMI criteria to detect Acute Coronary Occlusion (ACO) and poor inter-rater reliability. Emerging ECG features beyond the traditional STEMI criteria show promise in improving early ACO diagnosis, but complexity hinders widespread adoption. The potential integration of Artificial Neural Networks (ANN) holds promise for enhancing diagnostic accuracy and addressing reliability issues in ECG interpretation for ACO symptoms.

METHODS

Ovid MEDLINE, CINAHL, EMBASE, Cochrane, PubMed and Scopus were searched from inception through to 8th of December 2023. A thorough search of the grey literature and reference lists of relevant articles was also performed to identify additional studies. Articles were included if they reported the use of ANN for ECG interpretation of Acute Coronary Syndrome in the Emergency Department patients.

RESULTS

The search yielded a total of 244 articles. After removing duplicates and excluding non-relevant articles, 14 remained for analysis. There was significant heterogeneity in the types of ANN models used and the outcomes assessed, making direct comparisons challenging. Nevertheless, ANN appeared to demonstrate higher accuracy than physician interpreters for the evaluated outcomes and this proved independent of both specialty and years of experience.

CONCLUSIONS

The interpretation of ECGs in patients with suspected ACS using ANN appears to be accurate and potentially superior when compared to human interpreters and computerised algorithms. This appears consistent across various ANN models and outcome variables. Future investigations should emphasise ANN interpretation of ECGs in patients with ACO, where rapid and accurate diagnosis can significantly benefit patients through timely access to reperfusion therapies.

Compared with physician overread, computer is less accurate but helpful in interpretation of electrocardiography for ST-segment elevation myocardial infarction

INTRODUCTION

Previous studies have demonstrated varying sensitivity and specificity of computer-interpreted electrocardiography (CIE) in identifying ST-segment elevation myocardial infarction (STEMI). This study aims to evaluate the accuracy of contemporary computer software in recognizing electrocardiography (ECG) signs characteristic of STEMI compared to emergency physician overread in clinical practice.

MATERIAL AND METHODS

In this retrospective observational single-center study, we reviewed the records of patients in the emergency department (ED) who underwent ECGs and troponin tests. Both the Philips DXL 16-Lead ECG. Algorithm and on-duty emergency physicians interpreted each standard 12‑lead ECG. The sensitivity and specificity of computer interpretation and physician overread ECGs for the definite diagnosis of STEMI were calculated and compared.

RESULTS

Among the 9340 patients included in the final analysis, 133 were definitively diagnosed with STEMI. When "computer-reported infarct or injury" was used as the indicator, the sensitivity was 87.2% (95% CI 80.3% to 92.4%) and the specificity was 86.2% (95% CI 85.5% to 86.9%). When "physician-overread STEMI" was used as the indicator, the sensitivity was 88.0% (95% CI 81.2% to 93.0%) and the specificity was 99.9% (95% CI 99.8% to 99.9%). The area under the receiver operating characteristic curve for physician-overread STEMI and computer-reported infarct or injury were 0.939 (95% CI 0.907 to 0.972) and 0.867 (95% CI 0.834 to 0.900), respectively.

CONCLUSIONS

This study reveals that while the sensitivity of the computer in recognizing ECG signs of STEMI is similar to that of physicians, physician overread of ECGs is more specific and, therefore, more accurate than CIE.

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.

Prehospital Electrocardiogram Transmission and Prehospital Scene Time: A Retrospective Cohort Study

Background: Telemedical transmission of prehospital electrocardiograms (ECGs) to a target clinic may improve clinical workflows and speed of intervention. However, whether ECG transmission delays prehospital workflows remains controversial. Therefore, we aimed to clarify whether ECG transmission prolongs prehospital scene time in patients diagnosed with acute coronary syndrome (ACS). Methods: We retrospectively included all patients diagnosed with ACS by prehospital emergency physicians from July 2016 to June 2019 at a single academic center. The primary endpoint was the effect of ECG transmission on prehospital scene time. The secondary endpoints were the effects of ECG diagnosis on prehospital scene time and quality of care. Multivariable regression was used to account for patients' age, physician specialty, completion of specialty training, and whether emergencies occurred throughout the day or night shifts as potential confounders. Results: A total of 1,106 cases were included, of which 154 ECG transmissions were performed. ECG transmission prolonged prehospital scene time by an average of 3 min: adjusted regression coefficient [95% confidence interval (95% CI)]: 3.24 (1.7-4.8), p < 0.001. Prehospital treatment time was not influenced by prehospital ECG-based diagnosis (ST-elevation myocardial infarction [STEMI] vs. non-ST-elevation ACS [NSTE-ACS]): adjusted regression coefficient (95% CI): 0.7 (-1.3 to 2.7), p = 0.490. Emergency physicians adhered to local standard operating procedures in 739 of 1,007 (73%) patients diagnosed with NSTE-ACS and 93 of 99 (94%) patients diagnosed with STEMI. A STEMI diagnosis compared with NSTE-ACS was associated with five times higher odds of adhering to standard operating procedures (odds ratio [95% CI]: 5.6 [2.7-14.6], p < 0.001). Conclusion: The observed delay of ∼3 min in the prehospital scene time by ECG transmission is clinically irrelevant. For patients prehospitally diagnosed with NSTE-ACS who do not meet STEMI criteria, adherence to standard operating procedures seems unjustifiably low and should be improved.

Does Inclusion of Emergency Medicine (EM) Residents in ECG Screening for STEMI Change the Time to Catheterization Lab Activation?

BACKGROUND

Emergency medicine physicians must rapidly obtain and interpret an electrocardiogram (ECG) to quickly identify life-threatening cardiac emergencies such as ST-elevation myocardial infarction (STEMI). Although ECG interpretation is a critical component of residency education, few high-powered studies exploring the accuracy of resident ECG interpretation exist.

OBJECTIVES

This study aims to evaluate whether or not the inclusion of Third Year Emergency Medicine Resident ECG interpretations is noninferior to attending-only ECG interpretations in regard to time to STEMI activation.

METHODS

This was a retrospective noninferiority study of STEMI activation times before and after the inclusion of Third Year Emergency Medicine Resident resident ECG interpretations into the workflow at an academic, urban tertiary care center between November 2020 and April 2022, excluding prehospital activations. The primary outcome was the proportion of successful STEMI activations initiated within 5 minutes of ECG completion. An absolute decrease of 10% between groups was chosen as the noninferiority margin.

RESULTS

In the attending-only group, 26 (66.7%) cases resulted in successful STEMI activations compared to 31 cases (77.5%) in the combined group. The proportion of successful STEMI activations did not differ with resident screening, X 2 = 1.15, P = 0.28. The absolute difference between groups' successful activations was an increase of 11%, which lies within the noninferiority margin (+11%, 95% confidence interval, -8.68% to 30.7%). Average times to STEMI activation in the attending-only and combined groups were 7.59 minutes (Standard Deviation [SD], 10.19) and 5.13 minutes (SD, 6.95), respectively. Average door-to-balloon times for those undergoing Percutaneous Coronary Intervention were 72.74 minutes (SD, 20.76) in the attending-only group and 89.90 minutes (SD, 67.74) in the combination group. Two sample t-test showed no statistically significant difference between the 2 groups for average time to STEMI activation (difference = 2.46 minutes, 95% CI, -1.46 to 6.38) and average door-to-balloon time (difference = 17.16, 95% CI, -39.73 to 5.41).

CONCLUSION

The inclusion of emergency medicine PGY-3 residents in the ECG screening workflow is noninferior to attending-only interpretation of ECGs with regard to STEMI activation time.

Does the Electrocardiogram Machine Interpretation Affect the Ability to Accurately Diagnose ST-Elevation Myocardial Infarction by Emergency Physicians?

INTRODUCTION

An ST-elevation myocardial infarction (STEMI) can portend significant morbidity and mortality to the patient and therefore must be rapidly diagnosed by an emergency medicine (EM) physician. The primary aim of this study is to determine whether EM physicians are more or less likely to accurately diagnose STEMI on an electrocardiogram (ECG) if they are blinded to the ECG machine interpretation as opposed to if they are provided the ECG machine interpretation.

METHODS

We performed a retrospective chart review of adult patients over 18 years of age admitted to our large, urban tertiary care center with a diagnosis of STEMI from January 1, 2016, to December 31, 2017. From these patients' charts, we selected 31 ECGs to create a quiz that was presented twice to a group of emergency physicians. The first quiz contained the 31 ECGs without the computer interpretations revealed. The second quiz, presented to the same physicians 2 weeks later, contained the same set of ECGs with the computer interpretations revealed. Physicians were asked "Based on the ECG above, is there a blocked coronary artery present causing a STEMI?"

RESULTS

Twenty-five EM physicians completed two 31-question ECG quizzes for a total of 1550 ECG interpretations. On the first quiz with computer interpretations blinded, the overall sensitivity in identifying a "true STEMI" was 67.2% with an overall accuracy of 65.6%. On the second quiz in which the ECG machine interpretation was revealed, the overall sensitivity was 66.4% with an accuracy of 65.8 % in correctly identifying a STEMI. The differences in sensitivity and accuracy were not statistically significant.

CONCLUSION

This study demonstrated no significant difference in physicians blinded versus those unblinded to computer interpretations of possible STEMI.

Deep learning augmented ECG analysis to identify biomarker-defined myocardial injury

Chest pain is a common clinical complaint for which myocardial injury is the primary concern and is associated with significant morbidity and mortality. To aid providers' decision-making, we aimed to analyze the electrocardiogram (ECG) using a deep convolutional neural network (CNN) to predict serum troponin I (TnI) from ECGs. We developed a CNN using 64,728 ECGs from 32,479 patients who underwent ECG within 2 h prior to a serum TnI laboratory result at the University of California, San Francisco (UCSF). In our primary analysis, we classified patients into groups of TnI < 0.02 or ≥ 0.02 µg/L using 12-lead ECGs. This was repeated with an alternative threshold of 1.0 µg/L and with single-lead ECG inputs. We also performed multiclass prediction for a set of serum troponin ranges. Finally, we tested the CNN in a cohort of patients selected for coronary angiography, including 3038 ECGs from 672 patients. Cohort patients were 49.0% female, 42.8% white, and 59.3% (19,283) never had a positive TnI value (≥ 0.02 µg/L). CNNs accurately predicted elevated TnI, both at a threshold of 0.02 µg/L (AUC = 0.783, 95% CI 0.780-0.786) and at a threshold of 1.0 µg/L (AUC = 0.802, 0.795-0.809). Models using single-lead ECG data achieved significantly lower accuracy, with AUCs ranging from 0.740 to 0.773 with variation by lead. Accuracy of the multi-class model was lower for intermediate TnI value-ranges. Our models performed similarly on the cohort of patients who underwent coronary angiography. Biomarker-defined myocardial injury can be predicted by CNNs from 12-lead and single-lead ECGs.

Can triage nurses accurately interpret the electrocardiogram in the emergency department to predict acute cardiovascular events? A prospective observational study

AIMS

The prompt recording of the electrocardiogram (ECG) and its correct interpretation is crucial to the management of patients who present to the emergency department (ED) with cardiovascular symptoms. Since triage nurses represent the first healthcare professionals evaluating the patient, improving their ability in interpreting the ECG could have a positive impact on clinical management. This real-world study investigates whether triage nurses can accurately interpret the ECG in patients presenting with cardiovascular symptoms.

DESIGN

Prospective, single-centre observational study conducted in a general ED of General Hospital of Merano in Italy.

METHODS

For all patients included, the triage nurses and the emergency physicians were asked to independently interpret and classify the ECGs answering to dichotomous questions. We correlated the interpretation of the ECG made by the triage nurses with the occurrence of acute cardiovascular events. The inter-rater agreement in ECG interpretation between physicians and triage nurses was evaluated with Cohen's kappa analysis.

RESULTS

Four hundred and ninety-one patients were included. The inter-rater agreement between triage nurses and physicians in classifying an ECG as abnormal was good. Patients who developed an acute cardiovascular event were 10.6% (52/491), and in 84.6% (44/52) of them, the nurse accurately classified the ECG as abnormal, with a sensitivity of 84.6% and a specificity of 43.5%.

CONCLUSION

Triage nurses have a moderate ability in identifying alterations in specific components of the ECG but a good ability in identifying patterns indicative of time-dependent conditions correlated with major acute cardiovascular events.

IMPACT FOR NURSING

Triage nurses can accurately interpret the ECG in the ED to identify patients at high risk of acute cardiovascular events.

REPORTING METHOD

The study was reported according to the STROBE guidelines.

NO PATIENT OR PUBLIC CONTRIBUTION

The study did not involve any patients during its conduction.

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.

Development and validation of deep learning ECG-based prediction of myocardial infarction in emergency department patients

Myocardial infarction diagnosis is a common challenge in the emergency department. In managed settings, deep learning-based models and especially convolutional deep models have shown promise in electrocardiogram (ECG) classification, but there is a lack of high-performing models for the diagnosis of myocardial infarction in real-world scenarios. We aimed to train and validate a deep learning model using ECGs to predict myocardial infarction in real-world emergency department patients. We studied emergency department patients in the Stockholm region between 2007 and 2016 that had an ECG obtained because of their presenting complaint. We developed a deep neural network based on convolutional layers similar to a residual network. Inputs to the model were ECG tracing, age, and sex; and outputs were the probabilities of three mutually exclusive classes: non-ST-elevation myocardial infarction (NSTEMI), ST-elevation myocardial infarction (STEMI), and control status, as registered in the SWEDEHEART and other registries. We used an ensemble of five models. Among 492,226 ECGs in 214,250 patients, 5,416 were recorded with an NSTEMI, 1,818 a STEMI, and 485,207 without a myocardial infarction. In a random test set, our model could discriminate STEMIs/NSTEMIs from controls with a C-statistic of 0.991/0.832 and had a Brier score of 0.001/0.008. The model obtained a similar performance in a temporally separated test set of the study sample, and achieved a C-statistic of 0.985 and a Brier score of 0.002 in discriminating STEMIs from controls in an external test set. We developed and validated a deep learning model with excellent performance in discriminating between control, STEMI, and NSTEMI on the presenting ECG of a real-world sample of the important population of all-comers to the emergency department. Hence, deep learning models for ECG decision support could be valuable in the emergency department.

Beyond the ST-segment in Occlusion Myocardial Infarction (OMI): Diagnosing the OMI-nous

The ST-segment elevation (STE) myocardial infarction (MI)/non-STEMI (NSTEMI) paradigm has been the central dogma of emergency cardiology for the last 30 years. Although it was a major breakthrough when it was first introduced, it is now one of the most important obstacles to the further progression of modern MI care. In this article, we trace why a disease with an established underlying pathology (acute coronary occlusion [ACO]) was unintentionally labeled with a surrogate electrocardiographic sign (STEMI/NSTEMI) instead of pathologic substrate itself (ACO-MI/non-ACO-MI or occlusion MI [OMI]/non-OMI [NOMI] for short), how this fundamental mistake caused important clinical consequences, and why we should change this paradigm with a better one, namely OMI/NOMI paradigm.

Two-dimensional ECG-based cardiac arrhythmia classification using DSE-ResNet

Electrocardiogram (ECG) is mostly used for the clinical diagnosis of cardiac arrhythmia due to its simplicity, non-invasiveness, and reliability. Recently, many models based on the deep neural networks have been applied to the automatic classification of cardiac arrhythmia with great success. However, most models independently extract the internal features of each lead in the 12-lead ECG during the training phase, resulting in a lack of inter-lead features. Here, we propose a general model based on the two-dimensional ECG and ResNet with detached squeeze-and-excitation modules (DSE-ResNet) to realize the automatic classification of normal rhythm and 8 cardiac arrhythmias. The original 12-lead ECG is spliced into a two-dimensional plane like a grayscale picture. DSE-ResNet is used to simultaneously extract the internal and inter-lead features of the two-dimensional ECG. Furthermore, an orthogonal experiment method is used to optimize the hyper-parameters of DSE-ResNet and a multi-model voting strategy is used to improve classification performance. Experimental results based on the test set of China Physiological Signal Challenge 2018 (CPSC2018) show that our model has average \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$F_1= 0.817$$\end{document}F1=0.817 for classifying normal rhythm and 8 cardiac arrhythmias. Meanwhile, compared with the state-of-art model in CPSC2018, our model achieved the best \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$F_1$$\end{document}F1 in 2 sub-abnormal types. This shows that the model based on the two-dimensional ECG and DSE-ResNet has advantage in detecting some cardiac arrhythmias and has the potential to be used as an auxiliary tool to help doctors perform cardiac arrhythmias analysis.

Exploring decision making 'noise' when interpreting the electrocardiogram in the context of cardiac cath lab activation

In this commentary paper, we discuss the use of the electrocardiogram to help clinicians make diagnostic and patient referral decisions in acute care settings. The paper discusses the factors that are likely to contribute to the variability and noise in the clinical decision making process for catheterization lab activation. These factors include the variable competence in reading ECGs, the intra/inter rater reliability, the lack of standard ECG training, the various ECG machine and filter settings, cognitive biases (such as automation bias which is the tendency to agree with the computer-aided diagnosis or AI diagnosis), the order of the information being received, tiredness or decision fatigue as well as ECG artefacts such as the signal noise or lead misplacement. We also discuss potential research questions and tools that could be used to mitigate this 'noise' and improve the quality of ECG based decision making.

Artificial intelligence versus physicians on interpretation of printed ECG images: Diagnostic performance of ST-elevation myocardial infarction on electrocardiography

BACKGROUND

Smartphone-based ECG analyzer using camera input can be useful as everyone have it. The purpose of this study was to evaluate whether such a system can outperform clinicians in detecting ST-elevation myocardial infarction (STEMI) regardless of image acquisition conditions.

METHODS

We retrospectively enrolled suspected STEMI patients in an emergency department from January to October 2021. A multifaceted cardiovascular assessment system (Quantitative ECG, QCG™) using ECG images to produce a quantitative score (QCG score, ranging from 0 to 100) was compared to human experts of 7 emergency physicians and 3 cardiologists. Voting scores (number of participants answering "yes" for STEMI) were calculated for comparison. The system's robustness was evaluated using an equivalence test where we prove its performance metric (area under the curve of the receiver operating characteristic curve, AUC-ROC) changes within a predetermined equivalence range (-0.01 to 0.01) in 6 different environments (A combination of three different smartphones and two image sources including computer screen and paper).

RESULTS

187 patients (96 STEMI, 51.3%) were analyzed. AUC-ROC of QCG score was 0.919 (0.880-0.957). AUC-ROCs of voting scores, 0.856 (0.799-0.913) for all clinicians, 0.843 (0.786-0.900) for emergency physicians, 0.817 (0.756-0.877) for cardiologists, and 0.848 (0.790-0.905) for high-performance group were significantly lower compared to that of QCG score. The change in AUC-ROC by image acquisition condition was negligible with a narrow confidence interval within -0.01 to 0.01 confirming the equivalence.

CONCLUSIONS

Image-based AI system can outperform clinicians in STEMI diagnosis and its performance was robust to change in image acquisition conditions.

ST-Segment Elevation Secondary to Spontaneous Pneumomediastinum in the Setting of COVID-19 Infection: A Case Report and Literature Review

A 45-year-old male presented with shortness of breath, cough,and chest discomfort. He reported positive test results for coronavirus disease 2019 (COVID-19) four days prior; this was confirmed by a second test administered at the hospital. Results of a chest CT, consistent with COVID-19 pneumonia, also revealed pneumomediastinum (PM). EKG showed ST elevations in the inferior leads with no reciprocal changes. Emergent cardiac catheterization showed that he had no stenosis in his major coronary arteries. His symptoms resolved after 25 days of hospitalization and the patient was ultimately discharged. This case highlights the importance of recognizing spontaneous PM as a complication of COVID-19 along with its uncommon presentation of ST elevation in order to prevent unnecessary invasive procedures.

Automatic Multichannel Electrocardiogram Record Classification Using XGBoost Fusion Model

There is an increasing demand for automatic classification of standard 12-lead electrocardiogram signals in the medical field. Considering that different channels and temporal segments of a feature map extracted from the 12-lead electrocardiogram record contribute differently to cardiac arrhythmia detection, and to the classification performance, we propose a 12-lead electrocardiogram signal automatic classification model based on model fusion (CBi-DF-XGBoost) to focus on representative features along both the spatial and temporal axes. The algorithm extracts local features through a convolutional neural network and then extracts temporal features through bi-directional long short-term memory. Finally, eXtreme Gradient Boosting (XGBoost) is used to fuse the 12-lead models and domain-specific features to obtain the classification results. The 5-fold cross-validation results show that in classifying nine categories of electrocardiogram signals, the macro-average accuracy of the fusion model is 0.968, the macro-average recall rate is 0.814, the macro-average precision is 0.857, the macro-average F1 score is 0.825, and the micro-average area under the curve is 0.919. Similar experiments with some common network structures and other advanced electrocardiogram classification algorithms show that the proposed model performs favourably against other counterparts in F1 score. We also conducted ablation studies to verify the effect of the complementary information from the 12 leads and the auxiliary information of domain-specific features on the classification performance of the model. We demonstrated the feasibility and effectiveness of the XGBoost-based fusion model to classify 12-lead electrocardiogram records into nine common heart rhythms. These findings may have clinical importance for the early diagnosis of arrhythmia and incite further research. In addition, the proposed multichannel feature fusion algorithm can be applied to other similar physiological signal analyses and processing.

Interobserver Variability Among Experienced Electrocardiogram Readers To Diagnose Acute Thrombotic Coronary Occlusion In Patients with Out of Hospital Cardiac Arrest: Impact of Metabolic Milieu and Angiographic Culprit

OBJECTIVES

We sought to evaluate interobserver concordance among experienced electrocardiogram (ECG) readers in predicting acute thrombotic coronary occlusion (ATCO) in the context of abnormal metabolic milieu (AMM) following resuscitated out of hospital cardiac arrest (OHCA).

METHODS

OHCA patients with initial shockable rhythm who underwent invasive coronary angiography (ICA) were included. AMM was defined as one of: pH < 7.1, lactate > 2 mmol/L, serum potassium < 2.8 or > 6.0 mEq/L. The initial ECG following ROSC but prior to ICA was adjudicated by 2 experienced readers using classic ST elevation myocardial infarction [STEMI] and expanded criteria and their combination to predict ATCO on ICA.

RESULTS

152 consecutive patients (mean age 58 years, 76% male) met inclusion criteria. AMM was present in 77%; and 42% had ATCO on ICA. Sensitivity, specificity, PPV, NPV using classic STEMI criteria were 50%, 98%, 94%, 72% (c-statistic 0.74); whereas for combined (STEMI + expanded) criteria they were 69%, 88%, 81%, 79% respectively (c-statistic 0.79). Inter-observer agreement (kappa) was 0.7 for classic STEMI criteria, and 0.66 for combined criteria. Agreement between readers was consistently higher when ATCO was absent and with NMM (kappa 0.78), but lower in AMM (kappa 0.6).

CONCLUSIONS

Despite experienced ECG readers, there was only modest overall concordance in predicting ATCO in the context of resuscitated OHCA. Significant interobserver variations were noted dependent on metabolic milieu and angiographic ATCO. These observations fundamentally question the role of the 12-lead ECG as primary triaging tool for early angiography among patients with OHCA.

Evolution of single-lead ECG for STEMI detection using a deep learning approach

BACKGROUND

While ST-Elevation Myocardial Infarction (STEMI) door-to-balloon times are often below 90 min, symptom to door times remain long at 2.5-h, due at least in part to a delay in diagnosis.

OBJECTIVES

To develop and validate a machine learning-guided algorithm which uses a single‑lead electrocardiogram (ECG) for STEMI detection to speed diagnosis.

METHODS

Data was extracted from the Latin America Telemedicine Infarct Network (LATIN), a population-based Acute Myocardial Infarction (AMI) program that provides care to patients in Brazil, Colombia, Mexico, and Argentina through telemedicine.

SAMPLE

the first dataset was comprised of 8511 ECGs that were used for various machine learning experiments to test our Deep Learning approach for STEMI diagnosis. The second dataset of 2542 confirmed STEMI diagnosis EKG records, including specific ischemic heart wall information (anterior, inferior, and lateral), was derived from the previous dataset to test the STEMI localization model. Preprocessing: Detection of QRS complexes by wavelet system, segmentation of each EKG record into individual heartbeats with fixed window of 0.4 s to the left and 0.9 s to the right of main. Training & Testing: 90% and 10% of the total dataset, respectively, were used for both models.

CLASSIFICATION

two 1-D convolutional neural networks were implemented, two classes were considered for first models (STEMI/Not-STEMI) and three classes for the second model (Anterior/Inferior/Lateral) each corresponding to the heart wall affected. These individual probabilities were aggregated to generate the final label for each model.

RESULTS

The single‑lead ECG strategy was able to provide an accuracy of 90.5% for STEMI detection with Lead V2, which also yielded the best results overall among individual leads. STEMI Localization model provided promising results for anterior and inferior wall STEMIs but remained suboptimal for Lateral STEMI.

CONCLUSIONS

An Artificial Intelligence-enhanced single‑lead ECG is a promising screening tool. This technology provides an autonomous and accurate STEMI diagnostic alternative that can be incorporated into wearable devices, potentially providing patients reliable means to seek treatment early and offers the potential to thereby improve STEMI outcomes in the long run.

Optimal ECG-lead selection increases generalizability of deep learning on ECG abnormality classification

Deep learning (DL) has achieved promising performance in detecting common abnormalities from the 12-lead electrocardiogram (ECG). However, diagnostic redundancy exists in the 12-lead ECG, which could impose a systematic overfitting on DL, causing poor generalization. We, therefore, hypothesized that finding an optimal lead subset of the 12-lead ECG to eliminate the redundancy would help improve the generalizability of DL-based models. In this study, we developed and evaluated a DL-based model that has a feature extraction stage, an ECG-lead subset selection stage and a decision-making stage to automatically interpret multiple common ECG abnormality types. The data analysed in this study consisted of 6877 12-lead ECG recordings from CPSC 2018 (labelled as normal rhythm or eight types of ECG abnormalities, split into training (approx. 80%), validation (approx. 10%) and test (approx. 10%) sets) and 3998 12-lead ECG recordings from PhysioNet/CinC 2020 (labelled as normal rhythm or four types of ECG abnormalities, used as external text set). The ECG-lead subset selection module was introduced within the proposed model to efficiently constrain model complexity. It detected an optimal 4-lead ECG subset consisting of leads II, aVR, V1 and V4. The proposed model using the optimal 4-lead subset significantly outperformed the model using the complete 12-lead ECG on the validation set and on the external test dataset. The results demonstrated that our proposed model successfully identified an optimal subset of 12-lead ECG; the resulting 4-lead ECG subset improves the generalizability of the DL model in ECG abnormality interpretation. This study provides an outlook on what channels are necessary to keep and which ones may be ignored when considering an automated detection system for cardiac ECG abnormalities. This article is part of the theme issue 'Advanced computation in cardiovascular physiology: new challenges and opportunities'.

Short duration Vectorcardiogram based inferior myocardial infarction detection: class and subject-oriented approach

Myocardial infarction (MI) happens when blood stops circulating to an explicit segment of the heart causing harm to the heart muscles. Vectorcardiography (VCG) is a technique of recording direction and magnitude of the signals that are produced by the heart in a 3-lead representation. In this work, we present a technique for detection of MI in the inferior portion of heart using short duration VCG signals. The raw signal was pre-processed using the median and Savitzky-Golay (SG) filter. The Stationary Wavelet Transform (SWT) was used for time-invariant decomposition of the signal followed by feature extraction. The selected features using minimum-redundancy-maximum-relevance (mRMR) based feature selection method were applied to the supervised classification methods. The efficacy of the proposed method was assessed under both class-oriented and a more real-life subject-oriented approach. An accuracy of 99.14 and 89.37% were achieved respectively. Results of the proposed technique are better than existing state-of-art methods and used VCG segment is shorter. Thus, a shorter segment and a high accuracy can be helpful in the automation of timely and reliable detection of MI. The satisfactory performance achieved in the subject-oriented approach shows reliability and applicability of the proposed technique.

A deep learning algorithm for detecting acute myocardial infarction

BACKGROUND

Delayed diagnosis or misdiagnosis of acute myocardial infarction (AMI) is not unusual in daily practice. Since a 12-lead electrocardiogram (ECG) is crucial for the detection of AMI, a systematic algorithm to strengthen ECG interpretation may have important implications for improving diagnosis.

AIMS

We aimed to develop a deep learning model (DLM) as a diagnostic support tool based on a 12-lead electrocardiogram.

METHODS

This retrospective cohort study included 1,051/697 ECGs from 737/287 coronary angiogram (CAG)-validated STEMI/NSTEMI patients and 140,336 ECGs from 76,775 non-AMI patients at the emergency department. The DLM was trained and validated in 80% and 20% of these ECGs. A human-machine competition was conducted. The area under the receiver operating characteristic curve (AUC), sensitivity, and specificity were used to evaluate the performance of the DLM.

RESULTS

The AUC of the DLM for STEMI detection was 0.976 in the human-machine competition, which was significantly better than that of the best physicians. Furthermore, the DLM independently demonstrated sufficient diagnostic capacity for STEMI detection (AUC=0.997; sensitivity, 98.4%; specificity, 96.9%). Regarding NSTEMI detection, the AUC of the combined DLM and conventional cardiac troponin I (cTnI) increased to 0.978, which was better than that of either the DLM (0.877) or cTnI (0.950).

CONCLUSIONS

The DLM may serve as a timely, objective and precise diagnostic decision support tool to assist emergency medical system-based networks and frontline physicians in detecting AMI and subsequently initiating reperfusion therapy.

Can an ECG performed during emergency department triage and interpreted as normal by computer analysis safely wait for clinician review until the time of patient assessment? A pilot study

INTRODUCTION

Electrocardiograms (ECGs) are frequently performed during patient triage in Emergency Departments (EDs). Emergency Physicians (EPs) are interrupted during other tasks to review ECGs. Critics believe this practice could lead to distraction with consequent medical error and decision fatigue. ECGs can be interpreted by computer software at the time of capture; some evidence exists to suggest that an ECG performed during ED triage with an immediate computer interpretation (ICI) of 'normal' will seldom contain information necessitating a change to triage management.

MATERIAL AND METHODS

All ED triage ECGs performed in the Royal Derby Hospital between 13th July 2017 and 12th July 2018 in patients without chest pain and with an ICI of 'normal' were identified through a database search. Forty were randomly selected and reviewed by two EPs (blinded to patient details, ICI and outcome) who were asked to identify those that required a change to triage management.

RESULTS

The study processes were feasible. At least one of the two EP reviewers felt that a change to triage management was required in 48% of cases (e.g. "review patient", "obtain blood gas", "review old ECGs"); they agreed on the need for change of management in 13% of cases. An ICI of normal had a NPV of 53% (95% CI 37-67%) for the need for a change to triage management based upon ECG findings. Inter-observer agreement was poor (kappa = 0.17).

CONCLUSIONS

Based on these results, ED triage ECGs should still be presented to EPs for immediate review regardless of the ICI. Inter-observer agreement between EPs was poor. Further research is required to link triage ECG interpretation, need for intervention and patient outcome.

Deep learning analysis of resting electrocardiograms for the detection of myocardial dysfunction, hypertrophy, and ischaemia: a systematic review

The aim of this review was to assess the evidence for deep learning (DL) analysis of resting electrocardiograms (ECGs) to predict structural cardiac pathologies such as left ventricular (LV) systolic dysfunction, myocardial hypertrophy, and ischaemic heart disease. A systematic literature search was conducted to identify published original articles on end-to-end DL analysis of resting ECG signals for the detection of structural cardiac pathologies. Studies were excluded if the ECG was acquired by ambulatory, stress, intracardiac, or implantable devices, and if the pathology of interest was arrhythmic in nature. After duplicate reviewers screened search results, 12 articles met the inclusion criteria and were included. Three articles used DL to detect LV systolic dysfunction, achieving an area under the curve (AUC) of 0.89-0.93 and an accuracy of 98%. One study used DL to detect LV hypertrophy, achieving an AUC of 0.87 and an accuracy of 87%. Six articles used DL to detect acute myocardial infarction, achieving an AUC of 0.88-1.00 and an accuracy of 83-99.9%. Two articles used DL to detect stable ischaemic heart disease, achieving an accuracy of 95-99.9%. Deep learning models, particularly those that used convolutional neural networks, outperformed rules-based models and other machine learning models. Deep learning is a promising technique to analyse resting ECG signals for the detection of structural cardiac pathologies, which has clinical applicability for more effective screening of asymptomatic populations and expedited diagnostic work-up of symptomatic patients at risk for cardiovascular disease.

Interpretable deep learning for automatic diagnosis of 12-lead electrocardiogram

Electrocardiogram (ECG) is a widely used reliable, non-invasive approach for cardiovascular disease diagnosis. With the rapid growth of ECG examinations and the insufficiency of cardiologists, accurate and automatic diagnosis of ECG signals has become a hot research topic. In this paper, we developed a deep neural network for automatic classification of cardiac arrhythmias from 12-lead ECG recordings. Experiments on a public 12-lead ECG dataset showed the effectiveness of our method. The proposed model achieved an average F1 score of 0.813. The deep model showed superior performance than 4 machine learning methods learned from extracted expert features. Besides, the deep models trained on single-lead ECGs produce lower performance than using all 12 leads simultaneously. The best-performing leads are lead I, aVR, and V5 among 12 leads. Finally, we employed the SHapley Additive exPlanations method to interpret the model's behavior at both the patient level and population level.

•

We develop a deep learning model for the automatic diagnosis of ECG

•

We present benchmark results of 12-lead ECG classification

•

We find out the top performance single lead in diagnosing ECGs

•

We employ the SHAP method to enhance clinical interpretability

STEMI: A transitional fossil in MI classification?

An important task in emergency cardiology is distinguishing patients with acute coronary occlusion (ACO), who will benefit from emergent reperfusion therapy, from those without ongoing myocyte loss who can be managed with medical therapy and for whom potentially harmful invasive interventions can be deferred. The electrocardiogram is critical in this process. Although the ST-segment elevation myocardial infarction (STEMI)/non-STEMI paradigm is well-established, with "STEMI" representing ACO, its evidence base is poor, and this can have dire consequences. The universally recommended STEMI criteria do not accurately diagnose ACO; in fact, they miss more than one-fourth of the patients with ACO, and also result in a substantial burden of unnecessary catheterization laboratory activations. We here discuss why we believe it is time to change the current STEMI/non-STEMI paradigm.

Automatic Triage of 12-Lead ECGs Using Deep Convolutional Neural Networks

BACKGROUND

The correct interpretation of the ECG is pivotal for the accurate diagnosis of many cardiac abnormalities, and conventional computerized interpretation has not been able to reach physician‐level accuracy in detecting (acute) cardiac abnormalities. This study aims to develop and validate a deep neural network for comprehensive automated ECG triage in daily practice.

METHODS AND RESULTS

We developed a 37‐layer convolutional residual deep neural network on a data set of free‐text physician‐annotated 12‐lead ECGs. The deep neural network was trained on a data set with 336.835 recordings from 142.040 patients and validated on an independent validation data set (n=984), annotated by a panel of 5 cardiologists electrophysiologists. The 12‐lead ECGs were acquired in all noncardiology departments of the University Medical Center Utrecht. The algorithm learned to classify these ECGs into the following 4 triage categories: normal, abnormal not acute, subacute, and acute. Discriminative performance is presented with overall and category‐specific concordance statistics, polytomous discrimination indexes, sensitivities, specificities, and positive and negative predictive values. The patients in the validation data set had a mean age of 60.4 years and 54.3% were men. The deep neural network showed excellent overall discrimination with an overall concordance statistic of 0.93 (95% CI, 0.92–0.95) and a polytomous discriminatory index of 0.83 (95% CI, 0.79–0.87).

CONCLUSIONS

This study demonstrates that an end‐to‐end deep neural network can be accurately trained on unstructured free‐text physician annotations and used to consistently triage 12‐lead ECGs. When further fine‐tuned with other clinical outcomes and externally validated in clinical practice, the demonstrated deep learning–based ECG interpretation can potentially improve time to treatment and decrease healthcare burden.

Comparing physicians and experienced advanced practice practitioners on the interpretation of electrocardiograms in the emergency department

BACKGROUND

Many patients present to emergency departments (ED) in U.S. for evaluation of acute coronary syndrome, and a rapid electrocardiogram (ECG) and interpretation are imperative for initial triage. A growing number of advanced practice practitioners (APP) (e.g. physician assistants, nurse practitioners) are assisting patient care in the ED.

PURPOSE

This study aims to compare the interpretation of ECGs by experienced APPs, each having 10 or more years of experience, with resident physicians and attending physicians.

PATIENTS AND METHODS

99 ED providers were stratified into attendings, residents at varying levels, and APPs were tested to interpret 36 ECGs from a database of ECGs initially interpreted to be ST elevation myocardial infarctions, of which 24 were determined to have a culprit lesion by coronary intervention.

RESULTS

Attending physicians were the most sensitive (0.86, 95% CI of 0.80 to 0.92) and specific (0.69, 95% Cl of 0.60 to 0.79) at interpreting ECGs, but APPs and physicians in their first year of practice out of residency were almost equally as sensitive [(0.82, 95% CI of 0.76 to 0.88) and (0.82, 95% CI of 0.76 to 0.88)] and specific [(0.62, 95% cl of 0.52 to 0.73) and (0.65, 95% Cl of 0.56 to 0.75)].

CONCLUSION

This study suggests the possibility of changing ED workflow where experienced APPs can be responsible for initial screening of an ECG, thus allowing fewer interruptions for ED physicians.

Detection and Classification of Cardiac Arrhythmias by a Challenge-Best Deep Learning Neural Network Model

Electrocardiograms (ECGs) are widely used to clinically detect cardiac arrhythmias (CAs). They are also being used to develop computer-assisted methods for heart disease diagnosis. We have developed a convolution neural network model to detect and classify CAs, using a large 12-lead ECG dataset (6,877 recordings) provided by the China Physiological Signal Challenge (CPSC) 2018. Our model, which was ranked first in the challenge competition, achieved a median overall F1-score of 0.84 for the nine-type CA classification of CPSC2018's hidden test set of 2,954 ECG recordings. Further analysis showed that concurrent CAs were adequately predictive for 476 patients with multiple types of CA diagnoses in the dataset. Using only single-lead data yielded a performance that was only slightly worse than using the full 12-lead data, with leads aVR and V1 being the most prominent. We extensively consider these results in the context of their agreement with and relevance to clinical observations.

•

Accurate AI diagnosis of cardiac arrhythmia on ECG data from 11 hospitals

•

Capable of diagnosing concurrent cardiac arrhythmias

•

An ensemble model combining 12- and 1-lead models ranked first in CPSC2018

•

aVR and V1 found to be the best-performing single leads

Performance of seven ECG interpretation programs in identifying arrhythmia and acute cardiovascular syndrome

BACKGROUND

No direct comparison of current electrocardiogram (ECG) interpretation programs exists.

OBJECTIVE

Assess the accuracy of ECG interpretation programs in detecting abnormal rhythms and flagging for priority review records with alterations secondary to acute coronary syndrome (ACS).

METHODS

More than 2,000 digital ECGs from hospitals and databases in Europe, USA, and Australia, were obtained from consecutive adult and pediatric patients and converted to 10 s analog samples that were replayed on seven electrocardiographs and classified by the manufacturers' interpretation programs. We assessed ability to distinguish sinus rhythm from non-sinus rhythm, identify atrial fibrillation/flutter and other abnormal rhythms, and accuracy in flagging results for priority review. If all seven programs' interpretation statements did not agree, cases were reviewed by experienced cardiologists.

RESULTS

All programs could distinguish well between sinus and non-sinus rhythms and could identify atrial fibrillation/flutter or other abnormal rhythms. However, false-positive rates varied from 2.1% to 5.5% for non-sinus rhythm, from 0.7% to 4.4% for atrial fibrillation/flutter, and from 1.5% to 3.0% for other abnormal rhythms. False-negative rates varied from 12.0% to 7.5%, 9.9% to 2.7%, and 55.9% to 30.5%, respectively. Flagging of ACS varied by a factor of 2.5 between programs. Physicians flagged more ECGs for prompt review, but also showed variance of around a factor of 2. False-negative values differed between programs by a factor of 2 but was high for all (>50%). Agreement between programs and majority reviewer decisions was 46-62%.

CONCLUSIONS

Automatic interpretations of rhythms and ACS differ between programs. Healthcare institutions should not rely on ECG software "critical result" flags alone to decide the ACS workflow.

Diagnostic accuracy of prehospital electrocardiograms interpreted remotely by emergency physicians in myocardial infarction patients

BACKGROUND

Prehospital 12‑lead electrocardiogram (ECG) is the most widely used screening tool for recognition of ST-segment elevation myocardial infarction (STEMI). However, prehospital diagnosis of STEMI based solely on ECGs can be challenging.

OBJECTIVES

To evaluate the ability of emergency department (ED) physicians to accurately interpret prehospital 12‑lead ECGs from a remote location.

METHODS

All suspected prehospital STEMI patients who were transported by EMS and underwent angiography between 2006 and 2014 were included. We reviewed prehospital ECGs and grouped them based on: 1) presence or absence of a culprit artery lesion following angiography; and 2) whether they met the 3rd Universal Definition of Myocardial Infarction. We also described characteristics of ECGs that were misinterpreted by ED physicians.

RESULTS

A total of 625 suspected STEMI cases were reviewed. Following angiography, 94% (590/625) of patients were found having a culprit artery lesion, while 6% (35/625) did not. Among these 35 patients, 24 had ECGs that mimicked STEMI criteria and 9 had non-ischemic signs. Upon ECG reinterpretation, 92% (577/625) had standard STEMI criteria while 8% (48/625) did not. Among these 48 patients, 35 had ischemic signs ECGs and 13 did not. Characteristics of misinterpreted ECGs included pericarditis, early repolarization, STE > 1 mm (1‑lead only), and negative T-wave.

CONCLUSIONS

Remote interpretation of prehospital 12‑lead ECGs by ED physicians was a useful diagnostic tool in this EMS system. Even if the rate of ECG misinterpretation is low, there is still room for ED physicians operating from a remote location to improve their ability to accurately diagnose STEMI patients.

A model for predicting emergency physician opinion of electrocardiogram tracing data quality

BACKGROUND

Limited work has established an objective measure of ECG quality that correlates with physician opinion of the study. We seek to establish a threshold of acceptable ECG data quality for the purpose of ruling out STEMI derived from emergency physician opinion.

METHODS

A panel of three emergency physicians rated 240 12-Lead ECGs as being acceptable or unacceptable data quality. Each lead of the ECG had the following measurements recorded: baseline wander, QRS signal amplitude, and artifact amplitude. A lasso regression technique was used to create the model.

RESULTS

The area under the curve for the model using all 36 elements is 1.0, indicating a perfect fit. A simplified model using 22 terms has an area under the curve of 0.994.

CONCLUSIONS

This study demonstrated that emergency physician opinion of ECG quality for the purpose of ruling out STEMI can be predicted through a regression model.

Sex Differences in Timeliness of Reperfusion in Young Patients With ST-Segment-Elevation Myocardial Infarction by Initial Electrocardiographic Characteristics

BACKGROUND

Young women with ST-segment-elevation myocardial infarction experience reperfusion delays more frequently than men. Our aim was to determine the electrocardiographic correlates of delay in reperfusion in young patients with ST-segment-elevation myocardial infarction.

METHODS AND RESULTS

We examined sex differences in initial electrocardiographic characteristics among 1359 patients with ST-segment-elevation myocardial infarction in a prospective, observational, cohort study (2008-2012) of 3501 patients with acute myocardial infarction, 18 to 55 years of age, as part of the VIRGO (Variation in Recovery: Role of Gender on Outcomes of Young AMI Patients) study at 103 US and 24 Spanish hospitals enrolling in a 2:1 ratio for women/men. We created a multivariable logistic regression model to assess the relationship between reperfusion delay (door-to-balloon time >90 or >120 minutes for transfer or door-to-needle time >30 minutes) and electrocardiographic characteristics, adjusting for sex, sociodemographic characteristics, and clinical characteristics at presentation. In our study (834 women and 525 men), women were more likely to exceed reperfusion time guidelines than men (42.4% versus 31.5%; P<0.01). In multivariable analyses, female sex persisted as an important factor in exceeding reperfusion guidelines after adjusting for electrocardiographic characteristics (odds ratio, 1.57; 95% CI, 1.15-2.15). Positive voltage criteria for left ventricular hypertrophy and absence of a prehospital ECG were positive predictors of reperfusion delay; and ST elevation in lateral leads was an inverse predictor of reperfusion delay.

CONCLUSIONS

Sex disparities in timeliness to reperfusion in young patients with ST-segment-elevation myocardial infarction persisted, despite adjusting for initial electrocardiographic characteristics. Left ventricular hypertrophy by voltage criteria and absence of prehospital ECG are strongly positively correlated and ST elevation in lateral leads is negatively correlated with reperfusion delay.

The utility of the triage electrocardiogram for the detection of ST-segment elevation myocardial infarction

INTRODUCTION

Current AHA/ACC guidelines on the management of ST-elevation myocardial infarction (STEMI) suggest that an ECG is indicated within 10minutes of arrival for patients arriving to the Emergency Department (ED) with symptoms concerning for STEMI. In response, there has been a creep towards performing ECGs more frequently in triage. The objectives of this study were to quantify the number of triage ECGs performed at our institution, assess the proportion of ECGs performed within current hospital guidelines, and evaluate the rate of STEMI detection in triage ECGs.

METHODS

A retrospective chart review of all emergency department patients presenting over a period of 8days who had a triage ECG performed. Cases of bradycardia or tachycardia were excluded. Data collection included patient demographics, presenting complaint, cardiac risk factors, troponin values, and final diagnosis. Summary statistics are reported in a descriptive manner.

RESULTS

During the study period, 538 patients had a triage ECG for possible STEMI with no STEMI identified and 16 NSTEMI diagnoses (confirmed as positive troponins following ED assessment). Sixty-three (11.7%) patients did not meet internal criteria for a triage ECG. A NSTEMI ED diagnosis was identified in 3% of patients who met internal triage ECG criteria and 1.6% who did not meet criteria (p=0.29). A cost analysis was performed using an average of 50 STEMI cases diagnosed in our ED per given year. Current institutional ECG billing rates for ECGs performed and interpreted is $125 per ECG, providing an estimated triage ECG charge to detect one STEMI at $54,295.

DISCUSSION

This retrospective study of 538 triage ECG's performed over an 8day period identified no STEMIs and 16 NSTEMIs. A very large number of ECGs were done at triage overall and included patients who do not meet our own hospital criteria. Given the extremely low yield and high associated charges, current guidelines for triage ECG for identifying a possible STEMI should be reviewed.

Detection of STEMI Using Prehospital Serial 12-Lead Electrocardiograms

OBJECTIVE

Repeated or serial 12-lead electrocardiograms (ECGs) in the prehospital setting may improve management of patients with subtle ST-segment elevation (STE) or with a ST-segment elevation myocardial infarction (STEMI) that evolves over time. However, there is a minimal amount of scientific evidence available to support the clinical utility of this method. Our objective was to evaluate the use of serial 12-lead ECGs to detect STEMI in patients during transport in a Canadian emergency medical services (EMS) jurisdiction.

METHODS

We performed a retrospective study of suspected STEMI patients transported by EMS in the Chaudière-Appalaches region (Québec, Canada) between August 2006 and December 2013. Patients were monitored by a serial 12-lead ECG system where an averaged ECG was transmitted every 2 minutes. Following review by an emergency physician, ECGs were grouped as having either a persistent STE or a dynamic STE that evolved over time.

RESULTS

A total of 754 suspected STEMI patients were transported by EMS during the study period. Of these, 728 patients met eligibility criteria and were included in the analysis. A persistent STE was observed in 84.3% (614/728) of patients, while the remaining 15.7% (114/728) had a dynamic STE. Among those with dynamic STE, 11.1% (81/728) had 1 ST-segment change (41 no-STEMI to STEMI; 40 STEMI to no-STEMI) and 4.5% (33/728) had ≥ 2 ST-segment changes (17 no-STEMI to STEMI; 16 STEMI to no-STEMI). Overall, in 8.0% (58/728) of the cohort, STEMI was identified on a subsequent ECG following an initial no-STEMI ECG.

CONCLUSIONS

Through recognition of transient ST-segment changes during transport via the prehospital serial 12-lead ECG system, STEMI was identified in 8% of suspected STEMI patients who had an initial no-STEMI ECG. Key words: electrocardiography; emergency medical services; ST-elevation myocardial infarction; prehospital dynamic ECG.

New Insights Into the Use of the 12-Lead Electrocardiogram for Diagnosing Acute Myocardial Infarction in the Emergency Department

The 12-lead electrocardiogram (ECG) remains the most immediately accessible and widely used initial diagnostic tool for guiding management in patients with suspected myocardial infarction (MI). Although the development of high-sensitivity cardiac troponin assays has improved the rule-in and rule-out and risk stratification of acute MI without ST elevation, the immediate management of the subset of acute MI with acute coronary occlusion depends on integrating clinical presentation and ECG findings. Careful interpretation of the ECG might yield subtle features suggestive of ischemia that might facilitate more rapid triage of patients with subtle acute coronary occlusion or, conversely, in identification of ST-elevation MI mimics (pseudo ST-elevation MI patterns). Our goal in this review article is to consider recent advances in the use of the ECG to diagnose coronary occlusion MIs, including the application of rules that allow MI to be diagnosed on the basis of atypical ECG manifestations. Such rules include the modified Sgarbossa criteria allowing identification of acute MI in left bundle branch block or ventricular pacing, the 3- and 4-variable formula to differentiate normal ST elevation (formerly called early repolarization) from subtle ECG signs of left anterior descending coronary artery occlusion, the differentiation of ST elevation of left ventricular aneurysm from that of acute anterior MI, and the use of lead aVL in the recognition of inferior MI. Improved use of the ECG is essential to improving the diagnosis and appropriate early management of acute coronary occlusion MIs, which will lead to improved outcomes for patients who present with acute coronary syndrome.

A new 4-variable formula to differentiate normal variant ST segment elevation in V2-V4 (early repolarization) from subtle left anterior descending coronary occlusion - Adding QRS amplitude of V2 improves the model

INTRODUCTION

Precordial normal variant ST elevation (NV-STE), previously often called "early repolarization," may be difficult to differentiate from subtle ischemic STE due to left anterior descending (LAD) occlusion. We previously derived and validated a logistic regression formula that was far superior to STE alone for differentiating the two entities on the ECG. The tool uses R-wave amplitude in lead V4 (RAV4), ST elevation at 60 ms after the J-point in lead V3 (STE60V3) and the computerized Bazett-corrected QT interval (QTc-B). The 3-variable formula is: 1.196 x STE60V3 + 0.059 × QTc-B - 0.326 × RAV4 with a value ≥23.4 likely to be acute myocardial infarction (AMI).

HYPOTHESIS

Adding QRS voltage in V2 (QRSV2) would improve the accuracy of the formula.

METHODS

355 consecutive cases of proven LAD occlusion were reviewed, and those that were obvious ST elevation myocardial infarction were excluded. Exclusion was based on one straight or convex ST segment in V2-V6, 1 millimeter of summed inferior ST depression, any anterior ST depression, Q-waves, "terminal QRS distortion," or any ST elevation >5 mm. The NV-STE group comprised emergency department patients with chest pain who ruled out for AMI by serial troponins, had a cardiologist ECG read of "NV-STE," and had at least 1 mm of STE in V2 and V3. R-wave amplitude in lead V4 (RAV4), ST elevation at 60 ms after the J-point in lead V3 (STE60V3) and the computerized Bazett-corrected QT interval (QTc-B) had previously been measured in all ECGs; physicians blinded to outcome then measured QRSV2 in all ECGs. A 4-variable formula was derived to more accurately classify LAD occlusion vs. NV-STE and optimize area under the curve (AUC) and compared with the previous 3-variable formula.

RESULTS

There were 143 subtle LAD occlusions and 171 NV-STE. A low QRSV2 added diagnostic utility. The derived 4-variable formula is: 0.052*QTc-B - 0.151*QRSV2 - 0.268*RV4 + 1.062*STE60V3. The 3-variable formula had an AUC of 0.9538 vs. 0.9686 for the 4-variable formula (p = 0.0092). At the same specificity as the 3-variable formula [90.6%, at which cutpoint (≥23.4), 123 of 143 MI were correctly classified for 86% sensitivity], the sensitivity of the new formula at cutpoint ≥17.75 is 90.2%, with 129/143 correctly classified MI, identifying an additional 6 cases. The cutpoint with the highest accuracy (92.0%) was at a cutoff value ≥18.2, with 88.8% sensitivity, 94.7% specificity, and a positive and negative likelihood ratio of 16.9 (95% CI: 8.9-32) and 0.12 (95% CI: 0.07-0.19). At this cutpoint, it correctly classified an additional 11 cases (289 of 315, vs. 278 of 315): 127/143 for MI (an additional 4 cases) and 162/171 for NV-STE (an additional 7 cases).

CONCLUSION

On the ECG, a 4-variable formula was derived which adds QRSV2; it differentiates subtle LAD occlusion from NV-STE better than the 3-variable formula. At a value ≥18.2, the formula (0.052*QTc-B - 0.151*QRSV2 - 0.268*RV4 + 1.062*STE60V3) was very accurate, sensitive, and specific, with excellent positive and negative likelihood ratios. This formula needs to be validated.

Effectiveness of a Prehospital Wireless 12-Lead Electrocardiogram and Cardiac Catheterization Laboratory Activation for ST-Elevation Myocardial Infarction

The aim of the study was to determine the prevalence of false-positive and inappropriate cardiac catheterization laboratory (CCL) activation in patients suspected with ST-elevation myocardial infarction (STEMI) diverted to a percutaneous coronary intervention (PCI) facility after paramedics wireless 12-lead electrocardiogram transmission to an emergency physician at an online medical control center. This retrospective study collected data from medical records of patients with suspected STEMI from 2006 to 2014. It included demographics, coronaropathic risk factors, cardiac biomarkers, time from the first medical contact to treatment, and final diagnosis. Primary outcome was the rate of false-positive and inappropriate CCL activation. As secondary outcomes, we compared patient characteristics between cases of appropriate and inappropriate CCL activation, and we assessed the presence of cardiac biomarkers, time from first medical contact to start of PCI, and final diagnosis. Overall, 673 patients with suspected STEMI were included in the analysis. A total of 640 patients (95%) had coronarography, of which 10% (62 of 640) did not have a culprit coronary artery (false positive). Angiography was canceled for 5% (33 of 673) of patients. The total false-positive and inappropriate CCL activation rate was 14% (95 of 673). Average time from the first medical contact to the start of PCI was 47 ± 18.1 minutes. Unwanted CCL activations were more likely to involve men aged >65 years and patients with a history of coronary artery disease. In conclusion, our system of transmitted prehospital electrocardiography and STEMI interpretation by emergency physicians at an online medical control center showed a total false-positive and inappropriate CCL activation rate of 14% over the 8-year study period.

Exploring the Relationship Between Eye Movements and Electrocardiogram Interpretation Accuracy

Interpretation of electrocardiograms (ECGs) is a complex task involving visual inspection. This paper aims to improve understanding of how practitioners perceive ECGs, and determine whether visual behaviour can indicate differences in interpretation accuracy. A group of healthcare practitioners (n = 31) who interpret ECGs as part of their clinical role were shown 11 commonly encountered ECGs on a computer screen. The participants' eye movement data were recorded as they viewed the ECGs and attempted interpretation. The Jensen-Shannon distance was computed for the distance between two Markov chains, constructed from the transition matrices (visual shifts from and to ECG leads) of the correct and incorrect interpretation groups for each ECG. A permutation test was then used to compare this distance against 10,000 randomly shuffled groups made up of the same participants. The results demonstrated a statistically significant (α  0.05) result in 5 of the 11 stimuli demonstrating that the gaze shift between the ECG leads is different between the groups making correct and incorrect interpretations and therefore a factor in interpretation accuracy. The results shed further light on the relationship between visual behaviour and ECG interpretation accuracy, providing information that can be used to improve both human and automated interpretation approaches.

Validation of the modified Sgarbossa criteria for acute coronary occlusion in the setting of left bundle branch block: A retrospective case-control study

BACKGROUND

The modified Sgarbossa criteria were proposed in a derivation study to be superior to the original criteria for diagnosing acute coronary occlusion (ACO) in left bundle branch block (LBBB). The new rule replaces the third criterion (5 mm of excessively discordant ST elevation [STE]) with a proportion (at least 1 mm STE and STE/S wave ≤-0.25). We sought to validate the modified criteria.

METHODS

This retrospective case-control study was performed by chart review in 2 tertiary care center emergency departments (EDs) and 1 regional referral center. A billing database was used at 1 site to identify all ED patients with LBBB and ischemic symptoms between May 2009 and June 2012. In addition, all 3 sites identified LBBB ACO patients who underwent emergent catheterization. We measured QRS amplitude and J-point deviation in all leads, blinded to outcomes. Acute coronary occlusion was determined by angiographic findings and cardiac biomarker levels, which were collected blinded to electrocardiograms. Diagnostic statistics of each rule were calculated and compared using McNemar's test.

RESULTS

Our consecutive cohort search identified 258 patients: 9 had ACO, and 249 were controls. Among the 3 sites, an additional 36 cases of ACO were identified, for a total of 45 ACO cases and 249 controls. The modified criteria were significantly more sensitive than the original weighted criteria (80% vs 49%, P < .001) and unweighted criteria (80% vs 56%, P < .001). Specificity of the modified criteria was not statistically different from the original weighted criteria (99% vs 100%, P = .5) but was significantly greater than the original unweighted criteria (99% vs 94%, P = .004).

CONCLUSIONS

The modified Sgarbossa criteria were superior to the original criteria for identifying ACO in LBBB.

The impact of telemedicine in cardiac critical care

Telemedicine was recognized in the 1970s as a legitimate entity for applying the use of modern information and communications technologies to the delivery of health services. Telecardiology is one of the fastest growing fields in telemedicine. The advancement of technologies and Web-based applications has allowed better transmission of health care delivery. This article discusses current advancements, the scope of telemedicine in cardiology, and its application to the critically ill. The impact of telecardiology consultation continues to evolve and includes many promising applications with potential positive implications for admission rates, morbidity, and mortality.

Diagnostic accuracy of ST-segment elevation myocardial infarction by various healthcare providers

BACKGROUND

This study aimed to compare the accuracy of ECG interpretation for diagnosis of STEMI by different groups of healthcare professionals involved in the STEMI program at our institution.

METHODS

We selected 21 ECGs from patients with typical symptoms of MI that were diagnosed with STEMI, and 10 ECGs of STEMI mimics. STEMI mimic ECGs were repeated in the package with a story of typical and atypical chest pain. ECGs were interpreted to diagnose STEMI and identify need for initiation of the cardiac catheterization lab (CCL). Participants identified confidence in STEMI recognition, and average number of ECGs read per week.

RESULTS

A total of 64 participants completed the study package. Cardiologists were more likely to provide correct interpretation compared to other groups. False positive diagnoses were more likely made by paramedics when compared to cardiologists (p < 0.01). There was a positive correlation between increased exposure to ECGs and accurate STEMI diagnosis (r = 0.482, p < 0.001). A threshold of ≥ 20 ECGs read per week showed a statistically significant improvement in accuracy (p < 0.001). Self-reported confidence correlated positively with accuracy (r = 0.402, p =< 0.001). Changing the ECG narrative of the STEMI mimic ECGs had a significant effect on interpretation between groups (p = 0.043).

CONCLUSIONS

Our study showed that healthcare profession and number of ECGs reviewed per week are predictive of the accuracy of ECG interpretation of STEMI. Cardiologists are the most accurate diagnosticians, and are the least likely to falsely activate the CCL. Weekly exposure of ≥ 20 ECGs may improve diagnostic accuracy regardless of underlying experience.