Diagnostic performance of a computer-based ECG rhythm algorithm

Bridging ECG learning with emerging technologies: Advancing clinical excellence

As ECG technology rapidly evolves to improve patient care, accurate ECG interpretation will continue to be foundational for maintaining high clinical standards. Recent studies have exposed significant educational gaps, with many healthcare professionals lacking sufficient training and proficiency. Furthermore, integrating new software and hardware ECG technologies poses challenges about potential knowledge and skill erosion. This underscores the need for clinicians who are adept at integrating clinical expertise with technological proficiency. It also highlights the need for innovative solutions to enhance ECG interpretation among healthcare professionals in this rapidly evolving environment. This work explores the importance of aligning ECG education with technological advancements and proposes how this synergy could advance patient care in the future.

Artificial Intelligence for Cardiovascular Care-Part 1: Advances: JACC Review Topic of the Week

Recent artificial intelligence (AI) advancements in cardiovascular care offer potential enhancements in diagnosis, treatment, and outcomes. Innovations to date focus on automating measurements, enhancing image quality, and detecting diseases using novel methods. Applications span wearables, electrocardiograms, echocardiography, angiography, genetics, and more. AI models detect diseases from electrocardiograms at accuracy not previously achieved by technology or human experts, including reduced ejection fraction, valvular heart disease, and other cardiomyopathies. However, AI's unique characteristics necessitate rigorous validation by addressing training methods, real-world efficacy, equity concerns, and long-term reliability. Despite an exponentially growing number of studies in cardiovascular AI, trials showing improvement in outcomes remain lacking. A number are currently underway. Embracing this rapidly evolving technology while setting a high evaluation benchmark will be crucial for cardiology to leverage AI to enhance patient care and the provider experience.

Impact of Computer-Interpreted ECGs on the Accuracy of Healthcare Professionals

The interpretation of electrocardiograms (ECGs) involves a dynamic interplay between computerized ECG interpretation (CEI) software and human overread. However, the impact of computer ECG interpretation on the performance of healthcare professionals remains largely unexplored. The aim of this study was to evaluate the interpretation proficiency of various medical professional groups, with and without access to the CEI report. Healthcare professionals from diverse disciplines, training levels, and countries sequentially interpreted 60 standard 12-lead ECGs, demonstrating both urgent and nonurgent findings. The interpretation process consisted of 2 phases. In the first phase, participants interpreted 30 ECGs with clinical statements. In the second phase, the same 30 ECGs and clinical statements were randomized and accompanied by a CEI report. Diagnostic performance was evaluated based on interpretation accuracy, time per ECG (in seconds [s]), and self-reported confidence (rated 0 [not confident], 1 [somewhat confident], or 2 [confident]). A total of 892 participants from various medical professional groups participated in the study. This cohort included 44 (4.9%) primary care physicians, 123 (13.8%) cardiology fellows-in-training, 259 (29.0%) resident physicians, 137 (15.4%) medical students, 56 (6.3%) advanced practice providers, 82 (9.2%) nurses, and 191 (21.4%) allied health professionals. The inclusion of the CEI was associated with a significant improvement in interpretation accuracy by 15.1% (95% confidence interval, 14.3-16.0; P < 0.001), decrease in interpretation time by 52 s (-56 to -48; P < 0.001), and increase in confidence by 0.06 (0.03-0.09; P = 0.003). Improvement in interpretation accuracy was seen across all professional subgroups, including primary care physicians by 12.9% (9.4-16.3; P = 0.003), cardiology fellows-in-training by 10.9% (9.1-12.7; P < 0.001), resident physicians by 14.4% (13.0-15.8; P < 0.001), medical students by 19.9% (16.8-23.0; P < 0.001), advanced practice providers by 17.1% (13.3-21.0; P < 0.001), nurses by 16.2% (13.4-18.9; P < 0.001), allied health professionals by 15% (13.4-16.6; P < 0.001), physicians by 13.2% (12.2-14.3; P < 0.001), and nonphysicians by 15.6% (14.3-17.0; P < 0.001).CEI integration improves ECG interpretation accuracy, efficiency, and confidence among healthcare professionals.

Single-lead arrhythmia detection through machine learning: cross-sectional evaluation of a novel algorithm using real-world data

BACKGROUND

Computer-assisted interpretation of single-lead ECG is the preliminary method for clinicians to flag and further evaluate an arrhythmia of clinical importance for acutely ill patients. Critical scrutiny of novel detection algorithms is lacking, particularly in external real-world data sets. This study's objective was to evaluate a hybrid machine learning model's ability to classify eight arrhythmias from a single-lead ECG signal from acutely ill patients.

METHODS

This cross-sectional external retrospective evaluation of a previously trained hybrid machine learning model against an ECG reading team in the setting of home hospital care (acute care delivered at home substituting for traditional hospital care) draws from patients admitted at two hospitals in Boston, Massachusetts, USA between 12 June 2017 and 23 November 2019. We calculated classifier statistics for each arrhythmia, all arrhythmias and strips where the model identified normal sinus rhythm.

RESULTS

The model analysed 2 680 162 min of single-lead ECG data from 423 patients and identified 691 478 arrhythmias. Patients had a mean age of 70 years (SD, 18), 60% were female and 45% were white. For any arrhythmia, the model had a sensitivity of 98%, a specificity of 100%, an accuracy of 98%, a positive predictive value of 100%, a negative predictive value of 93% and an F1 Score of 99%. Performance was best for pause (F1 Score, 99%) and worst for paroxysmal supraventricular tachycardia (F1 Score, 92%). The model's false positive rate for any arrhythmia was 0.2%, ranging from 0.4% for pause to 7.2% for paroxysmal supraventricular tachycardia. The false negative rate for any arrhythmia was 1.9%.

CONCLUSIONS

A hybrid machine learning model was effective at classifying common cardiac arrhythmias from a single-lead ECG in real-world data.

Humanizing AI in medical training: ethical framework for responsible design

The increasing use of artificial intelligence (AI) in healthcare has brought about numerous ethical considerations that push for reflection. Humanizing AI in medical training is crucial to ensure that the design and deployment of its algorithms align with ethical principles and promote equitable healthcare outcomes for both medical practitioners trainees and patients. This perspective article provides an ethical framework for responsibly designing AI systems in medical training, drawing on our own past research in the fields of electrocardiogram interpretation training and e-health wearable devices. The article proposes five pillars of responsible design: transparency, fairness and justice, safety and wellbeing, accountability, and collaboration. The transparency pillar highlights the crucial role of maintaining the explainabilty of AI algorithms, while the fairness and justice pillar emphasizes on addressing biases in healthcare data and designing models that prioritize equitable medical training outcomes. The safety and wellbeing pillar however, emphasizes on the need to prioritize patient safety and wellbeing in AI model design whether it is for training or simulation purposes, and the accountability pillar calls for establishing clear lines of responsibility and liability for AI-derived decisions. Finally, the collaboration pillar emphasizes interdisciplinary collaboration among stakeholders, including physicians, data scientists, patients, and educators. The proposed framework thus provides a practical guide for designing and deploying AI in medicine generally, and in medical training specifically in a responsible and ethical manner.

Oral Class I and III antiarrhythmic drugs for maintaining sinus rhythm after catheter ablation of atrial fibrillation

BACKGROUND

Recurrence of atrial tachyarrhythmias (ATa) following catheter ablation for atrial fibrillation (AF) is a common problem. Antiarrhythmic drugs have been used shortly after ablation in an attempt to maintain sinus rhythm, particularly Class I and III agents. However, it still needs to be established if the use of Class I or III antiarrhythmic medications, or both, reduce the risk of recurrence of ATa.

OBJECTIVES

To assess the effects of oral Class I and III antiarrhythmic drugs versus control (standard medical therapy without Class I or III antiarrhythmics, or placebo) for maintaining sinus rhythm in people undergoing catheter ablation for AF.

SEARCH METHODS

We systematically searched CENTRAL, MEDLINE, Embase, Web of Science Core Collection, and two clinical trial registers without restrictions on language or date to 5 August 2022.

SELECTION CRITERIA

We sought published, unpublished, and ongoing parallel-design, randomised controlled trials (RCTs) involving adult participants undergoing ablation for AF, with subsequent comparison of Class I and/or III antiarrhythmic use versus control (standard medical therapy or non-Class I and/or III antiarrhythmic use).

DATA COLLECTION AND ANALYSIS

We used standard methodological procedures expected by Cochrane and performed meta-analyses with risk ratios (RR) and Peto odds ratios (Peto OR). Our primary outcomes were recurrence of atrial tachyarrhythmias; adverse events: thromboembolic events; adverse events: myocardial infarction; adverse events: new diagnosis of heart failure; and adverse events: requirement for one or more hospitalisations for atrial tachyarrhythmia. Our secondary outcomes were: all-cause mortality; and requirement for one or more repeat ablations. Where possible, we performed comparison analysis by Class I and/or III antiarrhythmic and divided follow-up periods for our primary outcome. We performed comprehensive assessments of risk of bias and certainty of evidence applying the GRADE methodology.

MAIN RESULTS

We included nine RCTs involving a total of 3269 participants. Participants were on average 59.3 years old; 71.0% were male; and 72.9% and 27.4% had paroxysmal and persistent AF, respectively. Class I and/or III antiarrhythmics may reduce recurrence of ATa at 0 to 3 months postablation (risk ratio (RR) 0.74, 95% confidence interval (CI) 0.59 to 0.94, 8 trials, 3046 participants, low-certainty evidence) and likely reduce recurrence at > 3 to 6 months, our a priori primary time point (RR 0.85, 95% CI 0.78 to 0.93, 5 trials, 2591 participants, moderate-certainty evidence). Beyond six months the evidence is very uncertain, and the benefit of antiarrhythmics may not persist (RR 1.14, 95% CI 0.84 to 1.55, 4 trials, 2244 participants, very low-certainty evidence). The evidence suggests that Class I and/or III antiarrhythmics may not increase the risk of thromboembolic events, myocardial infarction, all-cause mortality, or requirement for repeat ablation, at 0 to 3, > 3 to 6, and > 6 months (where data were available; low- to very low-certainty evidence). The use of Class I and/or III antiarrhythmics postablation likely reduces hospitalisations for ATa by approximately 57% at 0 to 3 months (RR 0.43, 95% CI 0.28 to 0.64, moderate-certainty evidence). No data were available beyond three months. No data were available on new diagnoses of heart failure. Fewer data were available for Class I and III antiarrhythmics individually. Based on only one and two trials (n = 125 to 309), Class I antiarrhythmics may have little effect on recurrence of ATa at 0 to 3, > 3 to 6, and > 6 months (RR 0.88, 95% CI 0.64 to 1.20, 2 trials, 309 participants; RR 0.54, 95% CI 0.25 to 1.19, 1 trial, 125 participants; RR 0.87, 95% CI 0.57 to 1.32, 1 trial, 125 participants; low-certainty evidence throughout); requirement for hospitalisation for ATa at 0 to 3 months (low-certainty evidence); or requirement for repeat ablation at 0 to 3 months (low-certainty evidence). No data were available for thromboembolic events, myocardial infarction, new diagnosis of heart failure, or all-cause mortality at any time points, or hospitalisation or repeat ablation beyond three months. Class III antiarrhythmics may have little effect on recurrence of ATa at up to 3 months and at > 3 to 6 months (RR 0.76, 95% CI 0.50 to 1.16, 4 trials, 599 participants, low-certainty evidence; RR 0.82, 95% CI 0.62 to 1.09, 2 trials, 318 participants, low-certainty evidence), and beyond 6 months one trial reported a possible increase in recurrence of ATa (RR 1.95, 95% CI 1.29 to 2.94, 1 trial, 112 participants, low-certainty evidence). Class III antiarrhythmics likely reduce hospitalisations for ATa at 0 to 3 months (RR 0.40, 95% CI 0.26 to 0.63, moderate-certainty evidence), and may have little effect on all-cause mortality (low- to very low-certainty evidence). The effect of Class III antiarrhythmics on thromboembolic events and requirement for repeat ablation was uncertain (very low-certainty evidence for both outcomes). No data were available for myocardial infarction or new diagnosis of heart failure at any time point, outcomes other than recurrence beyond 6 months, or for hospitalisation and repeat ablation > 3 to 6 months. We assessed the majority of included trials as at low or unclear risk of bias. One trial reported an error in the randomisation process, raising the potential risk of selection bias; most of the included trials were non-blinded; and two trials were at high risk of attrition bias.

AUTHORS' CONCLUSIONS

We found evidence to suggest that the use of Class I and/or III antiarrhythmics up to 3 months after ablation is associated with a reduced recurrence of ATa 0 to 6 months after ablation, which may not persist beyond 6 months, and an immediate reduction in hospitalisation for ATa 0 to 3 months after ablation. The evidence suggests there is no difference in rates of all-cause mortality, thromboembolic events, or myocardial infarction between Class I and/or III antiarrhythmics versus control.

Performance of a Convolutional Neural Network and Explainability Technique for 12-Lead Electrocardiogram Interpretation

Importance

Millions of clinicians rely daily on automated preliminary electrocardiogram (ECG) interpretation. Critical comparisons of machine learning-based automated analysis against clinically accepted standards of care are lacking.

Objective

To use readily available 12-lead ECG data to train and apply an explainability technique to a convolutional neural network (CNN) that achieves high performance against clinical standards of care.

Design, Setting, and Participants

This cross-sectional study was conducted using data from January 1, 2003, to December 31, 2018. Data were obtained in a commonly available 12-lead ECG format from a single-center tertiary care institution. All patients aged 18 years or older who received ECGs at the University of California, San Francisco, were included, yielding a total of 365 009 patients. Data were analyzed from January 1, 2019, to March 2, 2021.

Exposures

A CNN was trained to predict the presence of 38 diagnostic classes in 5 categories from 12-lead ECG data. A CNN explainability technique called LIME (Linear Interpretable Model-Agnostic Explanations) was used to visualize ECG segments contributing to CNN diagnoses.

Main Outcomes and Measures

Area under the receiver operating characteristic curve (AUC), sensitivity, and specificity were calculated for the CNN in the holdout test data set against cardiologist clinical diagnoses. For a second validation, 3 electrophysiologists provided consensus committee diagnoses against which the CNN, cardiologist clinical diagnosis, and MUSE (GE Healthcare) automated analysis performance was compared using the F1 score; AUC, sensitivity, and specificity were also calculated for the CNN against the consensus committee.

Results

A total of 992 748 ECGs from 365 009 adult patients (mean [SD] age, 56.2 [17.6] years; 183 600 women [50.3%]; and 175 277 White patients [48.0%]) were included in the analysis. In 91 440 test data set ECGs, the CNN demonstrated an AUC of at least 0.960 for 32 of 38 classes (84.2%). Against the consensus committee diagnoses, the CNN had higher frequency-weighted mean F1 scores than both cardiologists and MUSE in all 5 categories (CNN frequency-weighted F1 score for rhythm, 0.812; conduction, 0.729; chamber diagnosis, 0.598; infarct, 0.674; and other diagnosis, 0.875). For 32 of 38 classes (84.2%), the CNN had AUCs of at least 0.910 and demonstrated comparable F1 scores and higher sensitivity than cardiologists, except for atrial fibrillation (CNN F1 score, 0.847 vs cardiologist F1 score, 0.881), junctional rhythm (0.526 vs 0.727), premature ventricular complex (0.786 vs 0.800), and Wolff-Parkinson-White (0.800 vs 0.842). Compared with MUSE, the CNN had higher F1 scores for all classes except supraventricular tachycardia (CNN F1 score, 0.696 vs MUSE F1 score, 0.714). The LIME technique highlighted physiologically relevant ECG segments.

Conclusions and Relevance

The results of this cross-sectional study suggest that readily available ECG data can be used to train a CNN algorithm to achieve comparable performance to clinical cardiologists and exceed the performance of MUSE automated analysis for most diagnoses, with some exceptions. The LIME explainability technique applied to CNNs highlights physiologically relevant ECG segments that contribute to the CNN's diagnoses.

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.

An artificial intelligence-enabled ECG algorithm for comprehensive ECG interpretation: Can it pass the 'Turing test'?

Objective

To develop an artificial intelligence (AI)-enabled electrocardiogram (ECG) algorithm capable of comprehensive, human-like ECG interpretation and compare its diagnostic performance against conventional ECG interpretation methods.

Methods

We developed a novel AI-enabled ECG (AI-ECG) algorithm capable of complete 12-lead ECG interpretation. It was trained on nearly 2.5 million standard 12-lead ECGs from over 720,000 adult patients obtained at the Mayo Clinic ECG laboratory between 2007 and 2017. We then compared the need for human over-reading edits of the reports generated by the Marquette 12SL automated computer program, AI-ECG algorithm, and final clinical interpretations on 500 randomly selected ECGs from 500 patients. In a blinded fashion, 3 cardiac electrophysiologists adjudicated each interpretation as (1) ideal (ie, no changes needed), (2) acceptable (ie, minor edits needed), or (3) unacceptable (ie, major edits needed).

Results

Cardiologists determined that on average 202 (13.5%), 123 (8.2%), and 90 (6.0%) of the interpretations required major edits from the computer program, AI-ECG algorithm, and final clinical interpretations, respectively. They considered 958 (63.9%), 1058 (70.5%), and 1118 (74.5%) interpretations as ideal from the computer program, AI-ECG algorithm, and final clinical interpretations, respectively. They considered 340 (22.7%), 319 (21.3%), and 292 (19.5%) interpretations as acceptable from the computer program, AI-ECG algorithm, and final clinical interpretations, respectively.

Conclusion

An AI-ECG algorithm outperforms an existing standard automated computer program and better approximates expert over-read for comprehensive 12-lead ECG interpretation.

The association between body temperature and electrocardiographic parameters in normothermic healthy volunteers

Background

Previous studies reported that hypo‐ and hyperthermia are associated with several atrial and ventricular electrocardiographical parameters, including corrected QT (QTc) interval. Enhanced characterization of variations in QTc interval and normothermic body temperature aids in better understanding the underlying mechanism behind drug induced QTc interval effects. The analysis’ objective was to investigate associations between body temperature and electrocardiographical parameters in normothermic healthy volunteers.

Methods

Data from 3023 volunteers collected at our center were retrospectively analyzed. Subjects were considered healthy after review of collected data by a physician, including a normal tympanic body temperature (35.5‐37.5°C) and in sinus rhythm. A linear multivariate model with body temperature as a continuous was performed. Another multivariate analysis was performed with only the QT subintervals as independent variables and body temperature as dependent variable.

Results

Mean age was 33.8 ± 17.5 years and mean body temperature was 36.6 ± 0.4°C. Body temperature was independently associated with age (standardized coefficient [SC] = −0.255, P < .001), female gender (SC = +0.209, P < .001), heart rate (SC = +0.231, P < .001), P‐wave axis (SC = −0.051, P < .001), J‐point elevation in lead V4 (SC = −0.121, P < .001), and QTcF duration (SC = −0.061, P = .002). In contrast, other atrial and atrioventricular (AV) nodal parameters were not independently associated with body temperature. QT subinterval analysis revealed that only QRS duration (SC = −0.121, P < .001) was independently associated with body temperature.

Conclusion

Body temperature in normothermic healthy volunteers was associated with heart rate, P‐wave axis, J‐point amplitude in lead V4, and ventricular conductivity, the latter primarily through prolongation of the QRS duration.

A comprehensive artificial intelligence–enabled electrocardiogram interpretation program

Background

Automated computerized electrocardiogram (ECG) interpretation algorithms are designed to enhance physician ECG interpretation, minimize medical error, and expedite clinical workflow. However, the performance of current computer algorithms is notoriously inconsistent. We aimed to develop and validate an artificial intelligence–enabled ECG (AI-ECG) algorithm capable of comprehensive 12-lead ECG interpretation with accuracy comparable to practicing cardiologists.

Methods

We developed an AI-ECG algorithm using a convolutional neural network as a multilabel classifier capable of assessing 66 discrete, structured diagnostic ECG codes using the cardiologist’s final annotation as the gold-standard interpretation. We included 2,499,522 ECGs from 720,978 patients ≥18 years of age with a standard 12-lead ECG obtained at the Mayo Clinic ECG laboratory between 1993 and 2017. The total sample was randomly divided into training (n = 1,749,654), validation (n = 249,951), and testing (n = 499,917) datasets with a similar distribution of codes. We compared the AI-ECG algorithm’s performance to the cardiologist’s interpretation in the testing dataset using receiver operating characteristic (ROC) and precision recall (PR) curves.

Results

The model performed well for various rhythm, conduction, ischemia, waveform morphology, and secondary diagnoses codes with an area under the ROC curve of ≥0.98 for 62 of the 66 codes. PR metrics were used to assess model performance accounting for category imbalance and demonstrated a sensitivity ≥95% for all codes.

Conclusions

An AI-ECG algorithm demonstrates high diagnostic performance in comparison to reference cardiologist interpretation of a standard 12-lead ECG. The use of AI-ECG reading tools may permit scalability as ECG acquisition becomes more ubiquitous.

Artificial Intelligence-Enabled ECG: a Modern Lens on an Old Technology

PURPOSE OF REVIEW

To (i) review the concept of artificial intelligence (AI); (ii) summarize recent developments in artificial intelligence-enabled electrocardiogram (AI-ECG); (iii) address notable inherent limitations and challenges of AI-ECG; and (iv) discuss the future direction of the field.

RECENT FINDINGS

Advancements in machine learning and computing methods have led to application of AI-ECG and potential new applications to patient care. Further study is needed to verify previous findings in diverse populations as well as begin to confront the limitations needed for clinical implementation. Nearly one century after the Nobel Prize was awarded to Willem Einthoven for demonstrating that an electrocardiogram (ECG) could record the electrical signature of the heart, the ECG remains one of the most important diagnostic tests in modern medicine. We now stand at the edge of true ECG innovation. Simultaneous advancements in computing power, wireless technology, digitized data availability, and machine learning have led to the birth of AI-ECG algorithms with novel capabilities and real potential for clinical application. AI has the potential to improve diagnostic accuracy and efficiency by providing fully automated, unbiased, and unambiguous ECG analysis along with promising new findings that may unlock new value in the ECG. These breakthroughs may cause a paradigm shift in clinical workflow as well as patient monitoring and management.

Deep learning for comprehensive ECG annotation

BACKGROUND

Increasing utilization of long-term outpatient ambulatory electrocardiographic (ECG) monitoring continues to drive the need for improved ECG interpretation algorithms.

OBJECTIVE

The purpose of this study was to describe the BeatLogic® platform for ECG interpretation and to validate the platform using electrophysiologist-adjudicated real-world data and publicly available validation data.

METHODS

Deep learning models were trained to perform beat and rhythm detection/classification using ECGs collected with the Preventice BodyGuardian® Heart monitor. Training annotations were created by certified ECG technicians, and validation annotations were adjudicated by a team of board-certified electrophysiologists. Deep learning model classification results were used to generate contiguous annotation results, and performance was assessed in accordance with the EC57 standard.

RESULTS

On the real-world validation dataset, BeatLogic beat detection sensitivity and positive predictive value were 99.84% and 99.78%, respectively. Ventricular ectopic beat classification sensitivity and positive predictive value were 89.4% and 97.8%, respectively. Episode and duration F₁ scores (range 0–100) exceeded 70 for all 14 rhythms (including noise) that were evaluated. F₁ scores for 11 rhythms exceeded 80, 7 exceeded 90, and 5 including atrial fibrillation/flutter, ventricular tachycardia, ventricular bigeminy, ventricular trigeminy, and third-degree heart block exceeded 95.

CONCLUSION

The BeatLogic platform represents the next stage of advancement for algorithmic ECG interpretation. This comprehensive platform performs beat detection, beat classification, and rhythm detection/classification with greatly improved performance over the current state of the art, with comparable or improved performance over previously published algorithms that can accomplish only 1 of these 3 tasks.

Erroneous computer-based interpretations of atrial fibrillation and atrial flutter in a Swedish primary health care setting

Objective: To describe the incidence of incorrect computerized ECG interpretations of atrial fibrillation or atrial flutter in a Swedish primary care population, the rate of correction of computer misinterpretations, and the consequences of misdiagnosis.Design: Retrospective expert re-analysis of ECGs with a computer-suggested diagnosis of atrial fibrillation or atrial flutter.Setting: Primary health care in Region Kronoberg, Sweden.Subjects: All adult patients who had an ECG recorded between January 2016 and June 2016 with a computer statement including the words 'atrial fibrillation' or 'atrial flutter'.Main outcome measures: Number of incorrect computer interpretations of atrial fibrillation or atrial flutter; rate of correction by the interpreting primary care physician; consequences of misdiagnosis of atrial fibrillation or atrial flutter.Results: Among 988 ECGs with a computer diagnosis of atrial fibrillation or atrial flutter, 89 (9.0%) were incorrect, among which 36 were not corrected by the interpreting physician. In 12 cases, misdiagnosed atrial fibrillation/flutter led to inappropriate treatment with anticoagulant therapy. A larger proportion of atrial flutters, 27 out of 80 (34%), than atrial fibrillations, 62 out of 908 (7%), were incorrectly diagnosed by the computer.Conclusions: Among ECGs with a computer-based diagnosis of atrial fibrillation or atrial flutter, the diagnosis was incorrect in almost 10%. In almost half of the cases, the misdiagnosis was not corrected by the overreading primary-care physician. Twelve patients received inappropriate anticoagulant treatment as a result of misdiagnosis.Key pointsData regarding the incidence of misdiagnosed atrial fibrillation or atrial flutter in primary care are lacking. In a Swedish primary care setting, computer-based ECG interpretations of atrial fibrillation or atrial flutter were incorrect in 89 of 988 (9.0%) consecutive cases.Incorrect computer diagnoses of atrial fibrillation or atrial flutter were not corrected by the primary-care physician in 47% of cases.In 12 of the cases with an incorrect computer rhythm diagnosis, misdiagnosed atrial fibrillation or flutter led to inappropriate treatment with anticoagulant therapy.

Blood pressure-related electrocardiographic findings in healthy young individuals

Purpose: Elevated blood pressure induces electrocardiographic changes and is associated with an increase in cardiovascular disease later in life compared to normal blood pressure levels. The purpose of this study was to evaluate the association between normal to high normal blood pressure values (90-139/50-89 mmHg) and electrocardiographic parameters related to cardiac changes in hypertension in healthy young adults.Methods: Data from 1449 volunteers aged 18-30 years collected at our centre were analyzed. Only subjects considered healthy by a physician after review of collected data with systolic blood pressure values between 90 and 139 mmHg and diastolic blood pressure values between 50 and 89 mmHg were included. Subjects were divided into groups with 10 mmHg systolic blood pressure increment between groups for analysis of electrocardiographic differences. Backward multivariate regression analysis with systolic and diastolic blood pressure as a continuous variable was performed.Results: The mean age was 22.7 ± 3.0 years, 73.7% were male. P-wave area, ventricular activation time, QRS-duration, Sokolow-Lyon voltages, Cornell Product, J-point-T-peak duration corrected for heart rate and maximum T-wave duration were significantly different between systolic blood pressure groups. In the multivariate model with gender, body mass index and cholesterol, ventricular rate (standardized coefficient (SC): +0.182, p < .001), ventricular activation time in lead V6 (SC= +0.065, p = .048), Sokolow-Lyon voltage (SC= +0.135, p < .001), and Cornell product (SC= +0.137, p < .001) were independently associated with systolic blood pressure, while ventricular rate (SC= +0.179, p < .001), P-wave area in lead V1 (SC= +0.079, p = .020), and Cornell product (SC= +0.091, p = .006) were independently associated with diastolic blood pressure.Conclusion: Blood pressure-related electrocardiographic changes were observed incrementally in a healthy young population with blood pressure in the normal range. These changes were an increased ventricular rate, increased atrial surface area, ventricular activation time and increased ventricular hypertrophy indices on a standard 12 lead electrocardiogram.

A deep neural network for 12-lead electrocardiogram interpretation outperforms a conventional algorithm, and its physician overread, in the diagnosis of atrial fibrillation

Background

Automated electrocardiogram (ECG) interpretations may be erroneous, and lead to erroneous overreads, including for atrial fibrillation (AF). We compared the accuracy of the first version of a new deep neural network 12-Lead ECG algorithm (Cardiologs®) to the conventional Veritas algorithm in interpretation of AF.

Methods

24,123 consecutive 12-lead ECGs recorded over 6 months were interpreted by 1) the Veritas® algorithm, 2) physicians who overread Veritas® (Veritas® + physician), and 3) Cardiologs® algorithm. We randomly selected 500 out of 858 ECGs with a diagnosis of AF according to either algorithm, then compared the algorithms' interpretations, and Veritas® + physician, with expert interpretation. To assess sensitivity for AF, we analyzed a separate database of 1473 randomly selected ECGs interpreted by both algorithms and by blinded experts.

Results

Among the 500 ECGs selected, 399 had a final classification of AF; 101 (20.2%) had ≥1 false positive automated interpretation. Accuracy of Cardiologs® (91.2%; CI: 82.4–94.4) was higher than Veritas® (80.2%; CI: 76.5–83.5) (p < 0.0001), and equal to Veritas® + physician (90.0%, CI:87.1–92.3) (p = 0.12). When Veritas® was incorrect, accuracy of Veritas® + physician was only 62% (CI 52–71); among those ECGs, Cardiologs® accuracy was 90% (CI: 82–94; p < 0.0001). The second database had 39 AF cases; sensitivity was 92% vs. 87% (p = 0.46) and specificity was 99.5% vs. 98.7% (p = 0.03) for Cardiologs® and Veritas® respectively.

Conclusion

Cardiologs® 12-lead ECG algorithm improves the interpretation of atrial fibrillation.

Deep Learning Approaches to Detect Atrial Fibrillation Using Photoplethysmographic Signals: Algorithms Development Study

Background

Wearable devices have evolved as screening tools for atrial fibrillation (AF). A photoplethysmographic (PPG) AF detection algorithm was developed and applied to a convenient smartphone-based device with good accuracy. However, patients with paroxysmal AF frequently exhibit premature atrial complexes (PACs), which result in poor unmanned AF detection, mainly because of rule-based or handcrafted machine learning techniques that are limited in terms of diagnostic accuracy and reliability.

Objective

This study aimed to develop deep learning (DL) classifiers using PPG data to detect AF from the sinus rhythm (SR) in the presence of PACs after successful cardioversion.

Methods

We examined 75 patients with AF who underwent successful elective direct-current cardioversion (DCC). Electrocardiogram and pulse oximetry data over a 15-min period were obtained before and after DCC and labeled as AF or SR. A 1-dimensional convolutional neural network (1D-CNN) and recurrent neural network (RNN) were chosen as the 2 DL architectures. The PAC indicator estimated the burden of PACs on the PPG dataset. We defined a metric called the confidence level (CL) of AF or SR diagnosis and compared the CLs of true and false diagnoses. We also compared the diagnostic performance of 1D-CNN and RNN with previously developed AF detectors (support vector machine with root-mean-square of successive difference of RR intervals and Shannon entropy, autocorrelation, and ensemble by combining 2 previous methods) using 10 5-fold cross-validation processes.

Results

Among the 14,298 training samples containing PPG data, 7157 samples were obtained during the post-DCC period. The PAC indicator estimated 29.79% (2132/7157) of post-DCC samples had PACs. The diagnostic accuracy of AF versus SR was 99.32% (70,925/71,410) versus 95.85% (68,602/71,570) in 1D-CNN and 98.27% (70,176/71,410) versus 96.04% (68,736/71,570) in RNN methods. The area under receiver operating characteristic curves of the 2 DL classifiers was 0.998 (95% CI 0.995-1.000) for 1D-CNN and 0.996 (95% CI 0.993-0.998) for RNN, which were significantly higher than other AF detectors (P<.001). If we assumed that the dataset could emulate a sufficient number of patients in training, both DL classifiers improved their diagnostic performances even further especially for the samples with a high burden of PACs. The average CLs for true versus false classification were 98.56% versus 78.75% for 1D-CNN and 98.37% versus 82.57% for RNN (P<.001 for all cases).

Conclusions

New DL classifiers could detect AF using PPG monitoring signals with high diagnostic accuracy even with frequent PACs and could outperform previously developed AF detectors. Although diagnostic performance decreased as the burden of PACs increased, performance improved when samples from more patients were trained. Moreover, the reliability of the diagnosis could be indicated by the CL. Wearable devices sensing PPG signals with DL classifiers should be validated as tools to screen for AF.

Body mass index related electrocardiographic findings in healthy young individuals with a normal body mass index

INTRODUCTION

An increased body mass index (BMI) (>25 kg/m²) is associated with a wide range of electrocardiographic changes. However, the association between electrocardiographic changes and BMI in healthy young individuals with a normal BMI (18.5-25 kg/m²) is unknown. The aim of this study was to evaluate the association between BMI and electrocardiographic parameters.

METHODS

Data from 1,290 volunteers aged 18 to 30 years collected at our centre were analysed. Only subjects considered healthy by a physician after review of collected data with a normal BMI and in sinus rhythm were included in the analysis. Subjects with a normal BMI (18.5-25 kg/m²) were divided into BMI quartiles analysis and a backward multivariate regression analysis with a normal BMI as a continuous variable was performed.

RESULTS

Mean age was 22.7 ± 3.0 years, mean BMI was 22.0, and 73.4% were male. There were significant differences between the BMI quartiles in terms of maximum P-wave duration, P-wave balance, total P-wave area in lead V1, PR-interval duration, and heart axis. In the multivariate model maximum P-wave duration (standardised coefficient (SC) = +0.112, P < 0.001), P-wave balance in lead V1 (SC = +0.072, P < 0.001), heart axis (SC = -0.164, P < 0.001), and Sokolow-Lyon voltage (SC = -0.097, P < 0.001) were independently associated with BMI.

CONCLUSION

Increased BMI was related with discrete electrocardiographic alterations including an increased P-wave duration, increased P-wave balance, a leftward shift of the heart axis, and decreased Sokolow-Lyon voltage on a standard twelve lead electrocardiogram in healthy young individuals with a normal BMI.

Comparing clinical judgment with the MySurgeryRisk algorithm for preoperative risk assessment: A pilot usability study

BACKGROUND

Major postoperative complications are associated with increased cost and mortality. The complexity of electronic health records overwhelms physicians' abilities to use the information for optimal and timely preoperative risk assessment. We hypothesized that data-driven, predictive-risk algorithms implemented in an intelligent decision-support platform simplify and augment physicians' risk assessments.

METHODS

This prospective, nonrandomized pilot study of 20 physicians at a quaternary academic medical center compared the usability and accuracy of preoperative risk assessment between physicians and MySurgeryRisk, a validated, machine-learning algorithm, using a simulated workflow for the real-time, intelligent decision-support platform. We used area under the receiver operating characteristic curve to compare the accuracy of physicians' risk assessment for six postoperative complications before and after interaction with the algorithm for 150 clinical cases.

RESULTS

The area under the receiver operating characteristic curve of the MySurgeryRisk algorithm ranged between 0.73 and 0.85 and was significantly better than physicians' initial risk assessments (area under the receiver operating characteristic curve between 0.47 and 0.69) for all postoperative complications except cardiovascular. After interaction with the algorithm, the physicians significantly improved their risk assessment for acute kidney injury and for an intensive care unit admission greater than 48 hours, resulting in a net improvement of reclassification of 12% and 16%, respectively. Physicians rated the algorithm as easy to use and useful.

CONCLUSION

Implementation of a validated, MySurgeryRisk computational algorithm for real-time predictive analytics with data derived from the electronic health records to augment physicians' decision-making is feasible and accepted by physicians. Early involvement of physicians as key stakeholders in both design and implementation of this technology will be crucial for its future success.

The Computerized ECG: Friend and Foe

Computerized interpretation of the electrocardiogram (ECG) began in the 1950s when conversion of its analog signal to digital form became available. Since then, automatic computer interpretations of the ECG have become routine, even at the point of care, by the addition of interpretive algorithms to portable ECG carts. Now, more than 100 million computerized ECG interpretations are recorded yearly in the United States. These interpretations have contributed to medical care by reducing physician reading time and accurately interpreting most normal ECGs. But errors do occur. The computer cannot be held responsible for misinterpretations due to recording errors, such as muscle artifacts or lead reversal. But, in many abnormal ECGs, the computer makes its own errors-sometimes critical-in its incorrect detection of arrhythmias, pacemakers, and myocardial infarctions. These errors require that all computerized statements be over-read by trained physicians who have the advantage of clinical context, unavailable to the computer. Together, the computer and over-readers now provide the most accurate ECG interpretations available.

Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network

Computerized electrocardiogram (ECG) interpretation plays a critical role in the clinical ECG workflow¹. Widely available digital ECG data and the algorithmic paradigm of deep learning² present an opportunity to substantially improve the accuracy and scalability of automated ECG analysis. However, a comprehensive evaluation of an end-to-end deep learning approach for ECG analysis across a wide variety of diagnostic classes has not been previously reported. Here, we develop a deep neural network (DNN) to classify 12 rhythm classes using 91,232 single-lead ECGs from 53,549 patients who used a single-lead ambulatory ECG monitoring device. When validated against an independent test dataset annotated by a consensus committee of board-certified practicing cardiologists, the DNN achieved an average area under the receiver operating characteristic curve (ROC) of 0.97. The average F₁ score, which is the harmonic mean of the positive predictive value and sensitivity, for the DNN (0.837) exceeded that of average cardiologists (0.780). With specificity fixed at the average specificity achieved by cardiologists, the sensitivity of the DNN exceeded the average cardiologist sensitivity for all rhythm classes. These findings demonstrate that an end-to-end deep learning approach can classify a broad range of distinct arrhythmias from single-lead ECGs with high diagnostic performance similar to that of cardiologists. If confirmed in clinical settings, this approach could reduce the rate of misdiagnosed computerized ECG interpretations and improve the efficiency of expert human ECG interpretation by accurately triaging or prioritizing the most urgent conditions.

Relative contribution of modifiable risk factors for incident atrial fibrillation in Hispanics, African Americans and non-Hispanic Whites

BACKGROUND

Contribution of modifiable risk factors for the risk of new onset atrial fibrillation (AF) in minority populations is poorly understood. Our objective was to compare the population attributable risk (PAR) of various risk factors for incident AF between Hispanic, African American and non-Hispanic Whites.

METHODS

An ECG/EMR database was interrogated for individuals free of AF for development of subsequent AF from 2000 to 2013. Cox regression analysis controlled for age > 65, male gender, body mass index > 40 kg/m², systolic blood pressure > 140 mm Hg, diabetes mellitus, heart failure, socioeconomic status less than the first percentile in New York State, and race/ethnicity. PAR was calculated as (prevalence of X) ∗ (HR - 1)/HR, where HR is the hazard ratio, and X is the risk factor.

RESULTS

47,722 persons free of AF (43% Hispanic, 37% Black and 20% White) were followed for subsequent incident AF. Hypertension in African Americans and Hispanics had a 7.93% and 7.66% greater PAR compared with non-Hispanics Whites. Similar findings existed for the presence of heart failure, with a higher PAR in non-Whites compared to Whites.

CONCLUSION

In conclusion, modifiable risk factors play an important role in the risk of incident AF. Higher PAR estimates in African Americans and Hispanics were observed for elevated systolic blood pressure and heart failure. Identification of these modifiable risk factors for atrial fibrillation in non-White minorities may assist in targeting better prevention therapies and planning from a public health perspective. No funding sources were used for this study.

Relation of Body Mass Index to Development of Atrial Fibrillation in Hispanics, Blacks, and Non-Hispanic Whites

No previous studies have examined the interaction between body mass index (BMI) and race/ethnicity with the risk of atrial fibrillation (AF). We retrospectively followed 48,323 persons free of AF (43% Hispanic, 37% black, and 20% white; median age 60 years) for subsequent incident AF (ascertained from electrocardiograms). BMI categories included very severely underweight (BMI <15 kg/m²), severely underweight (BMI 15.1 to 15.9 kg/m²), underweight (BMI 16 to 18.4 kg/m²), normal (BMI 18.5 to 24.9 kg/m²), overweight (BMI 25.0 to 29.9 kg/m²), moderately obese (BMI 30 to 34.9 kg/m²), severely obese (BMI 35 to 39.9 kg/m²), and very severely obese (BMI >40 kg/m²). Cox regression analysis controlled for baseline covariates: heart failure, gender, age, treatment for hypertension, diabetes, PR length, systolic blood pressure, left ventricular hypertrophy, socioeconomic status, use of β blockers, calcium channel blockers, and digoxin. Over a follow-up of 13 years, 4,744 AF cases occurred. BMI in units of 10 was associated with the development of AF (adjusted hazard ratio 1.088, 95% confidence interval 1.048 to 1.130, p <0.01). When stratified by race/ethnicity, non-Hispanic whites compared with blacks and Hispanics had a higher risk of developing AF, noted in those whom BMI classes were overweight to severely obese. In conclusion, our study demonstrates that there exists a relation between obesity and race/ethnicity for the development of AF. Non-Hispanic whites had a higher risk of developing AF compared with blacks and Hispanics.

Risk factors for QTc interval prolongation

PURPOSE

Prolongation of the QTc interval may result in Torsade de Pointes, a ventricular arrhythmia. Numerous risk factors for QTc interval prolongation have been described, including the use of certain drugs. In clinical practice, there is much debate about the management of the risks involved. In this study, we quantified the effect of these risk factors on the length of the QTc interval.

METHODS

We analyzed all ECGs that were taken during routine practice between January 2013 and October 2016 in the Spaarne Gasthuis, a general teaching hospital in the Netherlands. We collected laboratory values in the week before the ECG recording and the drugs prescribed. For the identification of risk factors, we used multilevel linear regression analysis to correct for multiple ECG recordings per patient.

RESULTS

We included 133,359 ECGs in our study, taken in 40,037 patients. Patients using one QT-prolonging drug had a 11.08 ms (95% CI 10.63-11.52; p < 0.001) longer QTc interval. Patients using two QT-prolonging drugs had a 3.04 ms (95% CI 2.06-4.02; p < 0.001) increase in the QTc interval compared to patients using one QT-prolonging drug. Women had a longer QTc interval compared to men (16.30 ms 95% CI 14.59-18.01; p < 0.001). The QTc interval increased with increasing age, but the difference between men and women diminished. Other independent risk factors that significantly prolonged the QTc interval with at least 10 ms were hypokalemia, hypocalcemia, and the use of loop diuretics.

CONCLUSION

We identified and quantified various risk factors for QTc interval prolongation.

Computer-Interpreted Electrocardiograms: Benefits and Limitations

Computerized interpretation of the electrocardiogram (CIE) was introduced to improve the correct interpretation of the electrocardiogram (ECG), facilitating health care decision making and reducing costs. Worldwide, millions of ECGs are recorded annually, with the majority automatically analyzed, followed by an immediate interpretation. Limitations in the diagnostic accuracy of CIE were soon recognized and still persist, despite ongoing improvement in ECG algorithms. Unfortunately, inexperienced physicians ordering the ECG may fail to recognize interpretation mistakes and accept the automated diagnosis without criticism. Clinical mismanagement may result, with the risk of exposing patients to useless investigations or potentially dangerous treatment. Consequently, CIE over-reading and confirmation by an experienced ECG reader are essential and are repeatedly recommended in published reports. Implementation of new ECG knowledge is also important. The current status of automated ECG interpretation is reviewed, with suggestions for improvement.

Socioeconomic status and the development of atrial fibrillation in Hispanics, African Americans and non-Hispanic whites

BACKGROUND

Atrial fibrillation (AF) is the most common arrhythmia and is associated with significant morbidity and mortality. Despite having a higher burden of traditional AF risk factors, African American and Hispanic minorities have a lower incidence of AF when compared to non-Hispanic whites, referred to as the "racial paradox."

HYPOTHESIS

Lower SES among Hispanics and African Americans may help to explain the lower incidence rates of AF compared to non-Hispanic whites.

METHODS

An electrocardiogram/electronic medical records database in New York State was interrogated for individuals free of AF for development of subsequent AF from 2000 to 2013. SES was assessed per zip code via a composite of 6 measures Z-scored to the New York State average. SES was reclassified into decile groups. Cox regression analysis controlling for all baseline differences was used to estimate the independent predictive ability of SES for AF.

RESULTS

We identified 48 631 persons (43% Hispanic, 37% African Americans, and 20% non-Hispanic white; mean age 59 years; mean follow-up of 3.2 years) of which 4556 AF cases occurred. Hispanics and African Americans had lower AF risk than whites in all SES deciles (P < 0.001 by log-rank test). Higher SES was borderline associated with lower AF risk (hazard ratio: 0.990, 95% confidence interval: 0.980-1.001, P = 0.061). P trend analysis was not significant by any race/ethnic group by SES deciles for AF.

CONCLUSIONS

Our study suggests that non-Hispanic whites were at higher risk for AF compared to nonwhites, and this was independent of SES.

Screening strategies for atrial fibrillation: a systematic review and cost-effectiveness analysis

Background

Atrial fibrillation (AF) is a common cardiac arrhythmia that increases the risk of thromboembolic events. Anticoagulation therapy to prevent AF-related stroke has been shown to be cost-effective. A national screening programme for AF may prevent AF-related events, but would involve a substantial investment of NHS resources.

Objectives

To conduct a systematic review of the diagnostic test accuracy (DTA) of screening tests for AF, update a systematic review of comparative studies evaluating screening strategies for AF, develop an economic model to compare the cost-effectiveness of different screening strategies and review observational studies of AF screening to provide inputs to the model.

Design

Systematic review, meta-analysis and cost-effectiveness analysis.

Setting

Primary care.

Participants

Adults.

Intervention

Screening strategies, defined by screening test, age at initial and final screens, screening interval and format of screening {systematic opportunistic screening [individuals offered screening if they consult with their general practitioner (GP)] or systematic population screening (when all eligible individuals are invited to screening)}.

Main outcome measures

Sensitivity, specificity and diagnostic odds ratios; the odds ratio of detecting new AF cases compared with no screening; and the mean incremental net benefit compared with no screening.

Review methods

Two reviewers screened the search results, extracted data and assessed the risk of bias. A DTA meta-analysis was perfomed, and a decision tree and Markov model was used to evaluate the cost-effectiveness of the screening strategies.

Results

Diagnostic test accuracy depended on the screening test and how it was interpreted. In general, the screening tests identified in our review had high sensitivity (> 0.9). Systematic population and systematic opportunistic screening strategies were found to be similarly effective, with an estimated 170 individuals needed to be screened to detect one additional AF case compared with no screening. Systematic opportunistic screening was more likely to be cost-effective than systematic population screening, as long as the uptake of opportunistic screening observed in randomised controlled trials translates to practice. Modified blood pressure monitors, photoplethysmography or nurse pulse palpation were more likely to be cost-effective than other screening tests. A screening strategy with an initial screening age of 65 years and repeated screens every 5 years until age 80 years was likely to be cost-effective, provided that compliance with treatment does not decline with increasing age.

Conclusions

A national screening programme for AF is likely to represent a cost-effective use of resources. Systematic opportunistic screening is more likely to be cost-effective than systematic population screening. Nurse pulse palpation or modified blood pressure monitors would be appropriate screening tests, with confirmation by diagnostic 12-lead electrocardiography interpreted by a trained GP, with referral to a specialist in the case of an unclear diagnosis. Implementation strategies to operationalise uptake of systematic opportunistic screening in primary care should accompany any screening recommendations.

Limitations

Many inputs for the economic model relied on a single trial [the Screening for Atrial Fibrillation in the Elderly (SAFE) study] and DTA results were based on a few studies at high risk of bias/of low applicability.

Future work

Comparative studies measuring long-term outcomes of screening strategies and DTA studies for new, emerging technologies and to replicate the results for photoplethysmography and GP interpretation of 12-lead electrocardiography in a screening population.

Study registration

This study is registered as PROSPERO CRD42014013739.

Funding

The National Institute for Health Research Health Technology Assessment programme.

A simple, wide bandwidth, biopotential amplifier to record pacemaker pulse waveform

Reliable detection of pacemaker pulses is getting more and more important in electrocardiography (ECG) diagnosis. Many studies recommend ECG amplifiers with higher bandwidth to prevent errors. In the past, few pilot studies showed that analysis of pacemaker pulses waveform can enhance diagnosis (eg, lead failure and fractured wire), but they were carried out with inadequate instrumentations for clinical practice. Typically, pacemaker pulses last hundreds of microseconds, edges of pulses elapse in few microseconds, and amplitude may exhibit large variations from few millivolts to volts. Pulse waveforms change often and depend on pacemaker type and programming. A simple, biopotential amplifier made of a few off-the-shelf components is proposed. The circuit fulfills specifications for biopotential amplifiers and offers a large bandwidth (~1 MHz). Therefore, it is able to accurately record time course of pacemaker pulses and allows highly accurate pulse detection and timing. Signals can be easily displayed and acquired by means of a standard, battery-powered oscilloscope. Pacemaker pulse vectorcardiography can be obtained by using two or more, wideband channels. Some exemplificative waveforms recorded during patient's periodic medical examination are reported. The proposed circuit offers simultaneous conventional ECG signal as an additional output.

Validation of the Framingham Heart Study and CHARGE-AF Risk Scores for Atrial Fibrillation in Hispanics, African-Americans, and Non-Hispanic Whites

A risk score for atrial fibrillation (AF) has been developed by the Framingham Heart Study and Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE)-AF consortium. However, validation of these risk scores in an inner-city population is uncertain. Thus, a validation model was built using the Framingham Risk Score for AF and CHARGE-AF covariates. An in and outpatient electrocardiographic database was interrogated from 2000 to 2013 for the development of AF. Patients were included if their age was >45 and <95 years, had <10-year follow-up, if their initial electrocardiogram was without AF, had ≥ 2 electrocardiograms, and declared a race and/or ethnicity as non-Hispanic white, African-American, or Hispanic. For the Framingham Heart Study, 49,599 patients met inclusion criteria, of which 4,860 developed AF. Discrimination analysis using area under the curve (AUC) for original risk equations: non-Hispanic white AUC = 0.712 (95% confidence interval [CI] 0.694 to 0.731), African-American AUC = 0.733 (95% CI 0.716 to 0.751), and Hispanic AUC = 0.740 (95% CI 0.723 to 0.757). For the CHARGE-AF, 45,571 patients met inclusion criteria, of which 4,512 developed AF. Non-Hispanic white AUC = 0.673 (95% CI 0.652 to 0.694), African-American AUC = 0.706 (95% CI 0.685 to 0.727), and Hispanic AUC = 0.711 (95% CI 0.691 to 0.732). Calibration analysis showed qualitative similarities between cohorts. In conclusion, this is the first study to validate both the Framingham Heart Study and CHARGE-AF risk scores in both a Hispanic and African-American cohort. All models predicted AF well across all race and ethnic cohorts.

Estimating the prevalence of atrial fibrillation in a general population using validated electronic health data

Background

The purpose of this study was to determine the prevalence of atrial fibrillation (AF) in the general population and to validate an administrative diagnosis register, ie, the National Patient Register (NPR), and an electrocardiography (ECG) database in estimating disease prevalence.

Methods

The study was conducted in a well defined region in northern Sweden (population n=75,945) which consists of one hospital and eleven primary health care centers. Subjects with AF were identified by searching the combined inpatient and outpatient International Classification of Diseases (ICD)-based NPR (ICD-10 code I48) and an ECG database with computer-interpreted AF from January 1, 2004 to December 31, 2010. All identified cases with AF were validated.

Results

AF was confirmed in 2,274 patients. The overall prevalence was 3.0% (3.4% in men and 2.6% in women). AF prevalence rose steadily with age, and was 16.8% in patients aged 75 years and older and 21.9% in patients 85 years and older. Of all patients with validated AF, the NPR identified 93.2%. The ECG database identified an additional 6.8%, of which 81% were over 70 years of age. According to the NPR, the proportion of false positives and false negatives was 3.5% and 6.8%, respectively. The corresponding figure for the ECG database was 11.3% and 9.2%, respectively.

Conclusion

Our study shows a high prevalence of AF, especially among the elderly. Searching the ECG database enhanced the detection of AF. The reliability of the NPR was high, with a relatively low proportion of false positives and negatives.

Prolonged QT interval diagnosis suppression by a widely used computerized ECG analysis system

BACKGROUND

Clinicians commonly rely on automated diagnostic interpretations for initial, point-of-care identification of ECG abnormalities. Our study goal was to investigate how one widely used computerized ECG analysis system performs in labeling prolongation of heart rate-corrected QT interval (QTc), an arrhythmia risk marker.

METHODS AND RESULTS

ECGs acquired in 2009-2010 from patients ≥18 years old within the University of Michigan Health System, analyzed by the Marquette 12SL ECG Analysis Program (GE Healthcare), and exhibiting sinus rhythms with heart rate <100 beats per minute and QRS duration <120 ms constituted our database. Of 97 046 study ECGs (48.2% from males), a prolonged 12SL-calculated QTc value (ie, ≥470 ms in females >60 years old, and ≥460 ms in other sex/age groups) was displayed in 16 235 (16.7%). Nonetheless, for only 7709 (47.5%) of these ECGs with prolonged QTc did the automated interpretation include an accompanying "Prolonged QT" diagnostic statement. Such prolonged QT under-reporting was manifest across all patient environments and reflected algorithmic suppression of the diagnosis, attributable to ECG waveform-based criteria, in 8526 (52.5%) ECGs with prolonged QTc. Of the latter ECGs with prolonged QT diagnosis suppression, the computer declared 3588 (42.1%) "Normal" despite QTc prolongation.

CONCLUSIONS

In evaluating an adult patient whose 12SL-interpreted ECG lacks a prolonged QT diagnostic statement (assuming sinus rhythm <100 beats per minute and QRS duration <120 ms), physicians should examine the actual QTc value displayed on the report before concluding that this parameter is normal. Assessment of the clinical impact of prolonged QT diagnosis suppression by ECG waveform-based criteria is warranted.

Accurate ECG diagnosis of atrial tachyarrhythmias using quantitative analysis: a prospective diagnostic and cost-effectiveness study

Quantitative ECG Analysis.

INTRODUCTION

Optimal atrial tachyarrhythmia management is facilitated by accurate electrocardiogram interpretation, yet typical atrial flutter (AFl) may present without sawtooth F-waves or RR regularity, and atrial fibrillation (AF) may be difficult to separate from atypical AFl or rapid focal atrial tachycardia (AT). We analyzed whether improved diagnostic accuracy using a validated analysis tool significantly impacts costs and patient care.

METHODS AND RESULTS

We performed a prospective, blinded, multicenter study using a novel quantitative computerized algorithm to identify atrial tachyarrhythmia mechanism from the surface ECG in patients referred for electrophysiology study (EPS). In 122 consecutive patients (age 60 ± 12 years) referred for EPS, 91 sustained atrial tachyarrhythmias were studied. ECGs were also interpreted by 9 physicians from 3 specialties for comparison and to allow healthcare system modeling. Diagnostic accuracy was compared to the diagnosis at EPS. A Markov model was used to estimate the impact of improved arrhythmia diagnosis. We found 13% of typical AFl ECGs had neither sawtooth flutter waves nor RR regularity, and were misdiagnosed by the majority of clinicians (0/6 correctly diagnosed by consensus visual interpretation) but correctly by quantitative analysis in 83% (5/6, P = 0.03). AF diagnosis was also improved through use of the algorithm (92%) versus visual interpretation (primary care: 76%, P < 0.01). Economically, we found that these improvements in diagnostic accuracy resulted in an average cost-savings of $1,303 and 0.007 quality-adjusted-life-years per patient.

CONCLUSIONS

Typical AFl and AF are frequently misdiagnosed using visual criteria. Quantitative analysis improves diagnostic accuracy and results in improved healthcare costs and patient outcomes.

Improved pacemaker pulse detection: clinical evaluation of a new high-bandwidth electrocardiographic system

BACKGROUND

This study compares the pacemaker pulse detection performance of the new high-bandwidth (hi-fi) electrocardiographic (ECG) acquisition system to a conventional system in a prospective clinical evaluation.

METHODS

Electrocardiograms from 88 subjects with implanted pacemakers were recorded using different pacemaker programmed outputs and with different noise levels. Each ECG was recorded simultaneously from both systems. A cardiologist independently confirmed the clinically relevant ECGs. The pacemaker pulse detection sensitivity and positive predictive value (PPV) of each system were computed. The efficacy of each system was evaluated using a z test.

RESULTS

For the independently confirmed reports, the hi-fi system was superior, with higher sensitivity (99.2% vs 83.2%, P < .0001) and higher PPV (100% vs 99.9%, P = .33), for the detection of pacemaker pulses.

CONCLUSION

In a large group of subjects with implanted pacemakers and even in noisy conditions, the new hi-fi system was shown to improve pacemaker pulse detection sensitivity and PPV.

Observations of pacemaker pulses in high-bandwidth electrocardiograms and Dower-estimated vectorcardiograms

BACKGROUND

Electronic pacemaker pulses are poorly reproduced in computerized electrocardiogram (ECG) tracings, impairing both automated and human interpretation. In this study, a high-bandwidth system is used to examine ECG and vectorcardiogram characteristics of pacemaker pulses.

METHODS

In 69 subjects with artificial pacemakers, electrocardiograms were recorded at 75,000 samples per second with a high-bandwidth ECG system (GE Healthcare, Milwaukee, WI). Vectorcardiograms, as estimated with the Dower transform, were examined.

RESULTS

Pulse loops in the vectorcardiogram consisted of distinct discharge and recharge waves, with an angle difference of 174° ± 10° (mean ± SD) in 3 dimensions. Atrial pulses were on average oriented anteriorly, superiorly, and to the left; right ventricular pulses were oriented posteriorly, superiorly, and to the right; and left ventricular pulses were oriented posteriorly, inferiorly, and to the right. Other details of pacemaker pulses could be readily observed.

CONCLUSIONS

The high-bandwidth ECG has the potential to improve interpretation of paced rhythms in computerized ECGs.

Accuracy of diagnosing atrial flutter and atrial fibrillation from a surface electrocardiogram by hospital physicians: analysis of data from internal medicine departments

INTRODUCTION

Atrial fibrillation (AF) and atrial flutter (AFL) are clinically and electrocardiographically similar. However, considering significant therapeutic differences, differentiation of these 2 arrhythmias is essential. Our aims were to evaluate the misdiagnosis rate among electrocardiograms (ECGs) interpreted as AF or AFL by internists and to describe the factors that could be responsible for the misinterpretation.

METHODS

We evaluated patients discharged with a diagnosis of AF or AFL from internal medicine wards of a tertiary referral center. The reanalysis of the ECGs was performed by 2 senior cardiologists (1 electrophysiologist), blinded to the primary analysis and patient's clinical data.

RESULTS

The ECGs of 44 of 268 (16%) patients were misinterpreted and consisted of: 25 (57%) AFL, 5 (11%) SVT, 7 (16%) sinus rhythm with premature atrial beats and 7 (16%) AF. The baseline diagnosis was correct in 212 of 246 (86%) for AF and 12 of 22 (55%) for AFL, P < 0.001. A significantly higher rate of AFL was misdiagnosed compared with AF [25 of 37 (68%) versus 7 of 219 (3%), respectively; P < 0.001], higher in atypical than typical AFL [16 of 20 (80%) versus 9 of 17 (53%), respectively; P = 0.07]. Reduced quality ECGs was found more often among the incorrectly than the correctly diagnosed ECGs (P < 0.001].

CONCLUSIONS

ECGs, interpreted as AF or AFL by internists, are often misdiagnosed. AFL was misdiagnosed more often than AF, with atypical more often than typical AFL. Consulting with a cardiologist and applying diagnostic criteria may reduce misdiagnosis.

Troponin is superior to electrocardiogram and creatinine kinase MB for predicting clinically significant myocardial injury after coronary artery bypass grafting

AIMS

Cardiac biomarkers are routinely elevated after uncomplicated cardiac surgery to levels considered diagnostic of myocardial infarction in ambulatory populations. We investigated the diagnostic power of electrocardiogram (ECG) and cardiac biomarker criteria to predict clinically relevant myocardial injury using benchmarks of mortality and increased hospital length of stay (HLOS) in patients undergoing coronary artery bypass graft (CABG) surgery.

METHODS AND RESULTS

Perioperative ECGs, creatinine kinase MB fraction, and cardiac troponin I (cTnI) were assessed in 545 primary CABG patients. None of the ECG criteria for myocardial injury predicted mortality or HLOS. However, post-operative day (POD) 1 cTnI levels independently predicted 5-year mortality (hazard ratio = 1.42; 95% CI 1.14-1.76 for each 10 microg/L increase; P = 0.009), while adjusting for baseline demographic characteristics and perioperative risk factors. Moreover, cTnI was the only biomarker that significantly improved the prediction of 5-year mortality estimated by the logistic Euroscore (P = 0.02). Furthermore, the predictive value of cTnI for 5-year mortality was replicated in a separately collected cohort of 1031 CABG patients using cardiac troponin T.

CONCLUSION

Electrocardiogram diagnosis of post-operative myocardial injury after CABG does not independently predict an increased risk of 5-year mortality or HLOS. Conversely, cTnI is independently associated with an increased risk of mortality and prolonged HLOS.

Electrocardiography screening for cardiotoxicity after modified Vaccinia Ankara vaccination

BACKGROUND

Symptomatic myopericarditis has been described after smallpox vaccination using replication-competent vaccinia strains.

METHODS

We examined the incidence of new electrocardiogram (ECG) abnormalities and evaluated the safety and immunogenicity related to vaccination. Volunteer subjects (n=90) aged 18 to 32 years were enrolled in a National Institutes of Health-sponsored phase I smallpox vaccination trial (Division of Microbiology and Infectious Diseases 02-017) and observed over a 26-week period after 2 injections of IMVAMUNE, Modified Vaccinia Ankara vaccine (Bavarian Nordic A/S, Copenhagen, DK), followed by scarification with Dryvax (Wyeth Laboratories, Marietta, Penn). Diagnostic computer-derived ECG statements were available to the clinical study team and compared with those of a board-certified cardiologist who independently read the ECG tracings.

RESULTS

Serial ECG tracings available for 89 of the subjects revealed new ST-segment abnormalities in 2.2% and new T-wave abnormalities in 15.7%; the majority (71.4%) resolved on subsequent tracings. Cardiologist over-read of computer statements resulted in frequent changes in readings, particularly negation of cardiac arrhythmias. A cardiology consultation was requested in 17 subjects for nonspecific cardiac symptoms or new abnormal ECG findings. Echocardiograms were performed in 12 of the 17 subjects and were normal except for 1 subject with possible myopericarditis after receiving Dryvax.

CONCLUSION

New minor ECG abnormalities are common in apparently young healthy volunteers considered for smallpox vaccination trials. Cardiologist over-read of computer-generated ECG statements in vaccine trials using ECG as a screening tool for safety can reduce false-positive computer-determined ECG diagnoses and the need for inappropriate cardiology referral and additional noninvasive testing.

Automatic ECG wave extraction in long-term recordings using Gaussian mesa function models and nonlinear probability estimators

This paper describes the automatic extraction of the P, Q, R, S and T waves of electrocardiographic recordings (ECGs), through the combined use of a new machine-learning algorithm termed generalized orthogonal forward regression (GOFR) and of a specific parameterized function termed Gaussian mesa function (GMF). GOFR breaks up the heartbeat signal into Gaussian mesa functions, in such a way that each wave is modeled by a single GMF; the model thus generated is easily interpretable by the physician. GOFR is an essential ingredient in a global procedure that locates the R wave after some simple pre-processing, extracts the characteristic shape of each heart beat, assigns P, Q, R, S and T labels through automatic classification, discriminates normal beats (NB) from abnormal beats (AB), and extracts features for diagnosis. The efficiency of the detection of the QRS complex, and of the discrimination of NB from AB, is assessed on the MIT and AHA databases; the labeling of the P and T wave is validated on the QTDB database.

Causes of errors in the electrocardiographic diagnosis of atrial fibrillation by physicians

BACKGROUND

The emphasis of most large studies has been placed on the treatment and prevention of atrial fibrillation (AF) and its complications. Little is known about the accuracy of physicians in the electrocardiographic (ECG) diagnosis of AF and the possible causes of the diagnostic errors.

METHODS

Over a period of 10 months, a total of 35508 ECGs (28356 patients) were overread in a 385-bed community hospital within 24 hours of the initial reading. Corrected ECGs were returned to the patient file. The gold standard for the final diagnosis was based on the consensus by the cardiologist readers.

RESULTS

In all, 35508 ECGs were reviewed. A total of 2809 cases of AF were studied. Incorrect diagnoses related to AF were found in 219 cases. Type I errors (overdiagnosis) occurred in 137 cases. Rhythms with irregular R-R intervals (sinus rhythm with premature atrial contractions and atrial tachycardia or flutter with variable atrioventricular conduction) were misdiagnosed as AF. The presence of low-amplitude atrial activity and/or baseline artifact significantly increased the likelihood of the erroneous diagnosis, whereas ventricular rates of 130 beats/min did not influence the rate of error. Type II errors (missed AF) occurred in 82 cases where AF was either missed or confused with atrial tachycardia/flutter. Finally, ventricular pacing significantly increased the likelihood of type II errors.

CONCLUSIONS

In our institution, about 900 ECGs are read each week and 5 of them carry a wrong interpretation related to AF. More attention to common sources of errors as reinforced by an ongoing quality improvement program may reduce the rate of mistakes and thus prevent serious consequences.

Accuracy of diagnosing atrial fibrillation on electrocardiogram by primary care practitioners and interpretative diagnostic software: analysis of data from screening for atrial fibrillation in the elderly (SAFE) trial

OBJECTIVE

To assess the accuracy of general practitioners, practice nurses, and interpretative software in the use of different types of electrocardiogram to diagnose atrial fibrillation.

DESIGN

Prospective comparison with reference standard of assessment of electrocardiograms by two independent specialists.

SETTING

49 general practices in central England.

PARTICIPANTS

2595 patients aged 65 or over screened for atrial fibrillation as part of the screening for atrial fibrillation in the elderly (SAFE) study; 49 general practitioners and 49 practice nurses.

INTERVENTIONS

All electrocardiograms were read with the Biolog interpretative software, and a random sample of 12 lead, limb lead, and single lead thoracic placement electrocardiograms were assessed by general practitioners and practice nurses independently of each other and of the Biolog assessment.

MAIN OUTCOME MEASURES

Sensitivity, specificity, and positive and negative predictive values.

RESULTS

General practitioners detected 79 out of 99 cases of atrial fibrillation on a 12 lead electrocardiogram (sensitivity 80%, 95% confidence interval 71% to 87%) and misinterpreted 114 out of 1355 cases of sinus rhythm as atrial fibrillation (specificity 92%, 90% to 93%). Practice nurses detected a similar proportion of cases of atrial fibrillation (sensitivity 77%, 67% to 85%), but had a lower specificity (85%, 83% to 87%). The interpretative software was significantly more accurate, with a specificity of 99%, but missed 36 of 215 cases of atrial fibrillation (sensitivity 83%). Combining general practitioners' interpretation with the interpretative software led to a sensitivity of 92% and a specificity of 91%. Use of limb lead or single lead thoracic placement electrocardiograms resulted in some loss of specificity.

CONCLUSIONS

Many primary care professionals cannot accurately detect atrial fibrillation on an electrocardiogram, and interpretative software is not sufficiently accurate to circumvent this problem, even when combined with interpretation by a general practitioner. Diagnosis of atrial fibrillation in the community needs to factor in the reading of electrocardiograms by appropriately trained people.

Errors in the computerized electrocardiogram interpretation of cardiac rhythm

BACKGROUND

More than 100 million computer-interpreted electrocardiograms (ECG-C) are obtained annually. However, there are few contemporary published data on the accuracy of cardiac rhythm interpretation by this method.

PURPOSE

The purpose of this study is to determine the accuracy of ECG-C rhythm interpretation in a typical patient population.

METHODS

We compared the ECG-C rhythm interpretation to that of 2 expert overreaders in 2112 randomly selected standard 12-lead ECGs.

RESULTS

The ECG-C correctly interpreted the rhythm in 1858 and incorrectly identified the rhythm in 254 (overall accuracy, 88.0%). Sinus rhythm was correctly interpreted in 95.0% of the ECGs (1666/1753) with this rhythm, whereas nonsinus rhythms were correctly interpreted with an accuracy of only 53.5% (192/359) (P < .0001). The ECG-C interpreted sinus rhythm with a sensitivity of 95% (confidence interval, 93.8-96.7), specificity of 66.3%, and positive predictive value of 93.2%. The ECG-C interpreted nonsinus rhythms with a sensitivity of 72%, (confidence interval, 68.7-73.7), a specificity of 93%, and a positive predictive value of 59.3%. Of the 254 ECGs that had incorrect rhythm interpretation, additional major errors were noted in 137 (54%).

CONCLUSIONS

The ECG-C demonstrates frequent errors in the interpretation of nonsinus rhythms. In addition, incorrect rhythm interpretation by the ECG-C was frequently further compounded by additional major inaccuracies. Expert overreading of the ECG remains important in clinical settings with a high percentage of nonsinus rhythms.

Electrocardiograms with pacemakers: accuracy of computer reading

OBJECTIVE

We analyzed the accuracy with which a computer algorithm reads electrocardiograms (ECGs) with electronic pacemakers (PMs).

METHODS

Electrocardiograms were screened for the presence of electronic pacing spikes. Computer-derived interpretations were compared with cardiologists' readings.

RESULTS

Computer-drawn interpretations required revision by cardiologists in 61.3% of cases. In 18.4% of cases, the ECG reading algorithm failed to recognize the presence of a PM. The misinterpretation of paced beats as intrinsic beats led to multiple secondary errors, including myocardial infarctions in varying localization. The most common error in computer reading was the failure to identify an underlying rhythm. This error caused frequent misidentification of the PM type, especially when the presence of normal sinus rhythm was not recognized in a tracing with a DDD PM tracking the atrial activity.

CONCLUSION

The increasing number of pacing devices, and the resulting number of ECGs with pacing spikes, mandates the refining of ECG reading algorithms. Improvement is especially needed in the recognition of the underlying rhythm, pacing spikes, and mode of pacing.