ECG Interpretation Research Articles

Personalized LSTM Models for ECG Lead Transformations Led to Fewer Diagnostic Errors Than Generalized Models: Deriving 12-Lead ECG from Lead II, V2, and V6.

The prevalence of chronic cardiovascular diseases (CVDs) has risen globally, nearly doubling from 1990 to 2019. ECG is a simple, non-invasive measurement that can help identify CVDs at an early and treatable stage. A multi-lead ECG, up to 15 leads in a wearable form factor, is desirable. We seek to derive multiple ECG leads from a select subset of leads so that the number of electrodes can be reduced in line with a patient-friendly wearable device. We further compare personalized derivations to generalized derivations. Long-Short Term Memory (LSTM) networks using Lead II, V2, and V6 as input are trained to obtain generalized models using Bayesian Optimization for hyperparameter tuning for all patients and personalized models for each patient by applying transfer learning to the generalized models. We compare quantitatively using error metrics Root Mean Square Error (RMSE), R2, and Pearson correlation (ρ). We compare qualitatively by matching ECG interpretations of board-certified cardiologists. ECG interpretations from personalized models, when corrected for an intra-observer variance, were identical to the original ECGs, whereas generalized models led to errors. Mean performance values for generalized and personalized models were (RMSE-74.31 µV, R2-72.05, ρ-0.88) and (RMSE-26.27 µV, R2-96.38, ρ-0.98), respectively. Diagnostic accuracy based on derived ECG is the most critical validation of ECG derivation methods. Personalized transformation should be sought to derive ECGs. Performing a personalized calibration step to wearable ECG systems and LSTM networks could yield ambulatory 15-lead ECGs with accuracy comparable to clinical ECGs.

Pediatric resident education in pediatric cardiology – A needs assessment

BackgroundDetermining the topics general pediatric residents are taught during an inpatient pediatric cardiology rotation must be selected based on numerous factors, including resident career goals, core topics for board exams, and educator preferences. ObjectivesThis study aims to determine the preferred core cardiology concepts of both learners and educators as well the preferred teaching modalities of general pediatric residents while on an inpatient pediatric cardiology rotation. MethodsAll eligible residents in the Children's National Hospital pediatric residency program who had completed a cardiology rotation as well as pediatric cardiology fellows and attendings were asked to complete a survey. Participation in the study was voluntary. Respondents were asked to answer questions related to the perceived effectiveness and applicability of the cardiology curriculum, specifically inquiring about methods of teaching, the level taught, and the utility of topics selected. ResultsForty-seven (44 %) of the 106 eligible participants completed the survey. Seventeen unique topics in pediatric cardiology were ranked. The highest rated topics included congenital heart disease, murmurs, cardiac physiology, ECG interpretation, and syncope/dizziness. Participants reported the highest satisfaction with whiteboard and bedside teaching. ConclusionsThis study demonstrates a strong correlation between what learners and educators perceive should be the topics taught to residents to prepare them for their intended future career goals. These data provide a framework, both in teaching modalities and core concepts, for continued development and reassessment of inpatient pediatric cardiology rotations to optimally prepare pediatric residents with a strong foundation in pediatric cardiology.

Enhancing Emergency ECG Interpretation: A Practical Checklist Developed from a Study Assessing Young Physicians’ Diagnostic Abilities

Enhancing Emergency ECG Interpretation: A Practical Checklist Developed from a Study Assessing Young Physicians’ Diagnostic Abilities

Competency in ECG Interpretation and Arrhythmias Management among Critical Care Nurses in Saudi Arabia: A Cross Sectional Study.

Electrographic interpretation skills are important for healthcare practitioners caring for patients in need of cardiac assessment. Competency in ECG interpretation skills is critical to determine any abnormalities and initiate the appropriate care required. The purpose of the study was to determine the level of competence in electrocardiographic interpretation and knowledge in arrhythmia management of nurses in critical care settings. A descriptive cross-sectional design was used. A convenience sample of 255 critical care nurses from 4 hospitals in the Al-Madinah Region in Saudi Arabia was used. A questionnaire was designed containing a participant's characteristics and 10 questions with electrocardiographic strips. A pilot test was carried out to evaluate the validity and reliability of the questionnaire. Descriptive and bivariate analyses were conducted using an independent t-test, one-way ANOVA, or bi-variate correlation tests, as appropriate. A statistical significance of p < 0.05 was assumed. Females comprised 87.5% of the sample, and the mean age of the sample was 32.1 (SD = 5.37) years. The majority of the participants (94.9%) had taken electrocardiographic interpretation training courses. The mean total score of correct answers of all 10 ECG strips was 6.45 (±2.54) for ECG interpretation and 4.76 (±2.52) for arrhythmia management. No significant differences were observed between ECG competency level and nursing experience or previous training. Nurses working in the ICU and CCU scored significantly higher than those working in ED. The electrocardiographic knowledge in ECG interpretation and arrhythmia management of critical care nurses is low. Therefore, improving critical care nurses' knowledge of ECGs, identification, and management of cardiac arrhythmias is essential.

Layman electrocardiographic screening using smartphone-based multiple‑lead ECG device in school children

BackgroundPre-partecipation ECG screening of large populations has a significant socioeconomic impact. Technological progress now allows for high-tech-low-cost ECG screening using validated smartphone-based devices capable of guiding to the correct performance of a 12‑lead ECG by layman with no medical background. MethodsWe enrolled 728 (364, 52% males) individuals, aged 12–13 years who underwent ECG screening with a smartphone 12‑lead ECG during school hours by layman volunteers. Correct electrodes placement was provided by a validated image-processing algorithm by the smartphone camera in the App. ECG interpretation was via a telecardiology platform and alterations classified following current standards. ResultsA total of 741 ECGs were recorded, of which 13(2%) were technically not interpretable. Mean PR, QRS and QTc were: 145 ± 22, 85 ± 19 and 387 ± 57 msec. No QTc prolongation was observed. Mean QRS axis was 15°; 26 (4%) patients presented an iRBB. T-wave inversion from V1-V3 was present in 145 (21%) subjects. Twenty-one(3%) patients were referred to second level examination: deep Q-waves in inferior leads in 12(1.6%), ventricular ectopics in 5(0.7%), anterior T-waves inversions V1-V4 in 3(0.4%); extreme right axis deviation in 1(0.3%). Second line investigations did not provide any definitive diagnosis. Total project costs (material equipment and human cost) was 14.460€, 19.51€ per individual. The potential net saving with respect to current pre-participation screening cost was 19%. ConclusionsLayman 12‑lead Smartphone-ECG population screening proved feasible and effective, with a rate of non-interpretable ECG of <5%. Potential cost-saving in ECG screening and recording was 19%, providing an appealing opportunity when large campaigns should be addressed also in developing countries.

Competency and confidence in ECG interpretation among medical students.

To assess competency and confidence in ECG interpretation in medical students across years of medical school and evaluate the associations of various factors, a curriculum change, and student confidence with ECG competency. Four hundred and fourteen (414) third- to sixth-year medical students participated in this cross-sectional study conducted in 2019 in the Hebrew University of Jerusalem, Israel. A voluntary response sample of participants answered a validated, web-based questionnaire, composed of eight ECG strips. Participants were also asked about confidence and sources for ECG education and exposure. Competency and confidence across medical school years were compared using the ANOVA and chi-square tests. Competency was low overall (mean score, SD (standard deviation) 3.23±1.81 out of 8), and higher in sixth-year students compared to third-, fourth- and fifth-year students (4.37±1.69 vs. 2.90±1.82, 2.90±1.54, 2.50±1.56, respectively, F(3,337)=24.425, p<0.0001). There was no difference between students before and after the curriculum change. Work experience in medicine was associated with competency (odds ratio (OR), 7.97; 95% confidence interval (CI), 4.03-15.77, p<0.0001). The reported confidence level was low (median 2 out of 5) and was found to be correlated with the total score achieved (r(332)=0.5, p<0.0001). Student competency was shown to be insufficient throughout medical school. Competency and confidence in ECG interpretation seem to be significantly improved by increased and repetitive exposure to ECG. Thus, strategies to facilitate better ECG skills should involve an extended focus on ECG in the undergraduate and graduate curricula and include competency-based educational programs.

Normal limits of electrocardiogram in Africans and their consequences on the prevalence of left ventricular hypertrophy in hypertensive individuals: Insights from the TAHES study

AimsTo determine normal limits for major ECG variables, and the electrocardiographic impact of hypertension, in a rural sub-Saharan African setting. MethodsThis cross-sectional study included adults aged ≥25 years from Tanvè Health Study (TAHES) cohort. ECG were recorded at rest at 25 mm/s using a standard 12‑lead device. Wave amplitudes and durations were measured. Corrected QT interval (QTc) was calculated using Bazett's formula. Sokolow-Lyon, Cornell and Peguero-Lo Presti criteria were determined to assess left ventricular hypertrophy (LVH). ResultsECG was recorded among 997 out of 1407 TAHES participants. After exclusion of subjects with hypertension or diabetes, normal limits, defined as the 2nd and 98th percentiles, were evaluated in 622 healthy participants (median: 37 years; 60.1% women). The following limits were established in men (women): heart rate: 50 to 100 (55 to 102) beats/min, P wave duration: 80 to 120 (80 to 120) ms, PR interval: 120 to 200 (120 to 200) ms, QTc: 315 to 470 (323 to 465) ms, QRS duration: 50 to 120 (50 to 110) ms. Upper limits (in millimeter) for the Sokolow-Lyon, Cornell and Peguero-Lo Presti for men (women) were 47 (38), 30 (22) and 39 (30), respectively, all above current reference limits. The prevalence of LVH in hypertensive subjects according to these criteria were lower than those estimated according to current LVH criteria. ConclusionThe normal limits of ECG variables determined in this African population differ from those in Caucasians, indicating that ethnicity must be considered in ECG interpretation.

Electrocardiographic Parameters Associated with Adverse Outcomes in Children with Cardiomyopathies.

Cardiomyopathies have a low prevalence in children and thus may lead to malignant ventricular arrhythmias or the progression of heart failure, resulting in death. In adults, the QRS-T angle derived from ECG has been associated with adverse outcomes in patients with hypertrophic and dilated cardiomyopathies. We aimed to assess the electrocardiographic parameters, including QRS-T angle, associated with adverse cardiac events in children with cardiomyopathies. Forty-two children with cardiomyopathies were included in this study: 19 with dilated cardiomyopathy, 17 with hypertrophic cardiomyopathy, and 6 with left ventricular non-compaction. Additionally, 19 control subjects were recruited. In terms of ECG parameters, the QRS-T angle was significantly greater among patients with adverse outcomes compared to patients without the end points of the study (133° vs. 65°, p < 0.001). On Kaplan−Meier survival curves, QRS-T angle > 120°, increased serum concentrations of NT-proBNP and troponin I levels as well as greater NYHA or Ross scale were associated with the greatest risk of unfavorable outcome. The QRS-T angle appears to be a valuable component of 12-lead ECG interpretation, and might be helpful in outlining patients with the greatest cardiovascular risk. Additionally, serum biomarkers such as NT-proBNP (p = 0.003) and troponin (p < 0.001) are useful in outlining patients with the worst survival.

Abstract 11252: Developing an Interactive Online Curriculum for Teaching Pediatric Electrocardiogram Interpretation

Introduction: Interpretation of pediatric ECGs is a specific skill set that requires familiarity with normal electrographic changes in the developing heart, as well as recognition of a wide range of dysrhythmias or conduction alterations secondary to structural lesions. General pediatricians should use ECGs for various pediatric emergencies, common outpatient concerns and sports participation clearance. Previous studies showed that pediatric trainees are neither comfortable nor proficient in reading ECGs. We developed an interactive self- learning program for teaching pediatric ECG interpretation. Methods: In collaboration between pediatric residents and cardiologists, we created the curriculum, identifying high-yield topics, ‘red flags’ and age- related normal variances that are important to recognize. We built a 5-10-minute teaching module for each topic using Adobe Captivate. The modules were available online on a shared server. Participants were asked to take pre- and post-participation surveys and interpretation skill tests to define knowledge gaps and to focus our curricular development. Results: The pilot curriculum comprised of 14 high-yield topics, general and pediatric- focused. The modules included real ECG samples with multiple-choice questions testing specific electrographic aberrations, followed by audiovisual explanations. The pre-participation survey and test were completed by 57 participants: pediatric residents (59%) and others. Most reported that they do not feel comfortable with distinguishing normal from abnormal ECGs (77.2%), nor recognizing arrhythmias (75.5%) or identifying signs of structural abnormalities (98.3%). The mean test score was 5.13 ± 1.8 out of 10 questions answered correctly, with no significant difference between pediatric residents or other trainees ( p =0.78). Conclusions: We present a newly developed educational program for ECG interpretation that centers on visual learning and pattern recognition of high yield findings in pediatrics. Initial surveys confirm the need for an accessible and engaging teaching tool and we have shared the online modules to be available outside of our institution. We intend to continue and develop the program as more feedback is received.

Abstract 10066: Clonal Hematopoiesis of Indeterminate Potential is Not Associated With Incident Atrial Fibrillation

Introduction: Inflammation is involved in the initiation and maintenance of atrial fibrillation (AF). Clonal hematopoiesis of indeterminate potential (CHIP) increases cardiovascular risk through pro-inflammatory mechanisms. Hypothesis: The presence of CHIP is associated with increased risk of incident AF. Methods: Individuals with AF were identified from the CATHGEN cohort, a study of 9,334 sequential individuals undergoing cardiac catheterization at Duke University Hospital, using ICD-9/10 codes and ECG interpretations. Incident AF was defined as AF diagnosed after blood sample collection. Individuals with predisposing or provoking conditions leading to AF (i.e., cardiomyopathy, valve disease, congenital heart disease, thyroid disease, and recent STEMI or cardiac surgery) were excluded. The remaining individuals with incident AF were matched 1:1 to controls without an AF diagnosis based on age, sex, ancestry-defined race, BMI, diabetes, hypertension, and smoking status. Whole exome sequencing was performed in DNA extracted from whole blood, and CHIP carriers in 68 genes of interest were identified. Results: Among 1,716 individuals with incident AF, 335 had no predisposing or provoking conditions leading to AF. Among these individuals, 17 (5.1%) were CHIP carriers, compared to 18 (5.4%) matched controls (odds ratio, 0.94; 95% confidence interval [CI], 0.48-1.86; p =0.9). Ten individuals (3.0%) with incident AF were carriers of a CHIP clone with variant allele fraction &gt;10%, compared to 11 (3.3%) matched controls (odds ratio 0.90; 95% CI, 0.37-2.22; p =0.8). Restriction to the two most common genes ( DNMT3A or TET2 ) also did not show an association between CHIP and AF ( Table 1 ). Conclusions: There is no association between CHIP carrier status, high burden of CHIP, or CHIP in DNMT3A or TET2 and incident AF in the CATHGEN cohort.

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.

Updates on Post-Resuscitation Care. After the Return of Spontaneous Circulation beyond the 2021 Guidelines.

Out-of-hospital cardiac arrest is one of the leading causes of mortality worldwide. The goal of resuscitation is often meant as the return of spontaneous circulation (ROSC). However, ROSC is only one of the steps towards survival. The post-ROSC phase is still a challenging one during which the risk of death is all but averted. Morbidity and mortality are exceedingly high due to cardiovascular and neurologic issues; for this reason, post ROSC care relies on international guidelines, the latest being published on April 2021. Since then, several studies have become available covering a variety of topics of crucial importance for post-resuscitation care such as the interpretation of the post-ROSC ECG, the timing of coronary angiography, the role of complete myocardial revascularization and targeted temperature management. This narrative review focuses on these new evidences, in order to further improve clinical practice, and on the need for a multidisciplinary and integrated system of care.

Pseudo-V1 ECG lead recording with quality smartwatches and medical interpretation may be necessary for timely and accurate remote detection of atrial flutter/fibrillation

Abstract Background Remote rhythm monitoring with wearable devices is increasingly used especially for early detection of atrial fibrillation/flutter (AF/Afl), being the access to hospital discouraged, especially for frail elderly patients, due to the burden and risk of COVID-19 pandemic. Whereas devices using photo plethysmography (PPG) may misinterpret as AF pulse irregularities due to extrasystoles, patient-directed recording of a single (usually wrist-to-wrist) lead ECG (LEAD I) with hand-held devices or smartwatches have been developed to increase accuracy in AF detection. However, although recent studies validating such devices single-lead ECG recording have shown high sensitivity and specificity, false negative findings such as those reported here are still possible and must be prevented [1]. Purpose Given previous experience of diagnostic uncertainty or failure of the smartwatch ECG (SW-ECG) LEAD I to detect AF/Afl, we have tested if false negative diagnosis could be avoided by recording in addition at least one right precordial (pseudo-V1) lead analyzed by a trained healthcare professional. Method Over one calendar year observation, five patients with previous history of ablated supraventricular arrhythmias suffering sudden palpitations suspected of paroxysmal AF/Afl were instructed to record with their smartwatch at least one precordial lead in addition to LEAD I, to monitor ECG until the termination of symptoms. The SW-ECG strips were sent by telephone for professional interpretation. Diagnostic accuracy based on LEAD I and pseudo-V1 were independently validated by two cardiologists (diagnostic goldstandard - DGS). Results 22 AF/Afl events occurred. Pharmacological cardioversion to sinus rhythm (SR) was obtained in 64%. 192 ECG strips were transmitted. 43,7% of the strips were automatically classified as “not significant” (or “not valid”). Compared to DGS, out of 108 “valid” strips, correct automatic identification of AF/Afl was obtained in 36,4% with LEAD I, in 33,3% with pseudo V1 and in 54,5% with combined leads, respectively. Interestingly, the SW algorithm has wrongly diagnosed as SR, not only LEAD I, but also 39,4% of pseudo-V1 strips, despite clear-cut evidence of typical flutter waves (Figure 1), when RR intervals were regular due to high degree (e.g., 4:1) A-V block. Conclusions With simple instructions, patients (or their relatives) can easily record an additional precordial (pseudo-V1) SW-ECG lead, that may enhance sensitivity and specificity for remote detection of AF/Afl. However, at present, visual interpretation of SW-ECG by a trained healthcare professional is still needed to guarantee 100% correct diagnosis of AF/Afl, crucial to reduce thromboembolic risk and timely initiate the appropriate treatments. The automatic interpretation of SW's ECG could be improved by appropriate training of a machine learning approach to detect and analyze the atrial waveform provided by an additional pseudo-V1 lead. Funding Acknowledgement Type of funding sources: None.

Electrocardiography Interpretation Proficiency Among Medical Doctors of Different Grades in the United Kingdom.

Electrocardiography interpretation is a core clinical skill for all doctors participating in emergency medical services. Given the majority ofECGsinitially performed in hospital on admission are reviewed by junior doctors, and the need for life-threatening pathology to be diagnosed or excluded, further understanding of the level of competency in interpretation and factors associated with this are needed. This was a cross-sectional, descriptive, analytical study. The data were collected using a structured questionnaire. This was comprised of two sections; the first section contained questions related to confidence, previous ECG learning and factors thought to be associated with ECG interpretation competence. The second section was an ECG quiz of 10 12-lead ECGs of varying complexity for interpretation assessment. Descriptive and inferential statistics were utilized for data analysis. Sixty-twodoctors from foundation year 1 to registrar level working in acute medicine across three hospitals participated. The mean overall percentage score for the ECG quiz was 45%. No association was found between junior doctor training grade and overall score on the ECG assessment. Undergraduate and postgraduate teaching strategies also did not impact competence. Only 9.7% reported themselves as "confident" interpreting ECGs. There was a trend towards higher levels of competency among those who felt they had undergone sufficient ECG teaching and those who sought regular feedback from other clinicians. This study demonstrated low overall levels of ECG interpretation competency among junior doctors in a large acute teaching NHS trust regardless of grade. Factors associated with competency remain unclear.

European Union Digital education framework: A quality standard to guide the design of healthcare apps

Digital Education Initiatives and Timely Solutions (DIG-IT) is an Erasmus+ project collaboration between university academics, clinical educators, industry partners, healthcare professionals, and technology experts over 5 European Union (EU) states. The objectives are to create digital educational capacity for academic faculty and mobile learning opportunities for continuing professional healthcare education. The first deliverable of this project was the innovative European Union Digital Education Quality Standards Framework and Toolkit or EU-DEF. This framework can be used to consider all principle and secondary aspects when designing, delivering, and evaluating digital education resources. In this article we explore the framework efficacy in collaboratively developing two open access, mobile healthcare apps, ECG Interpretation for Nurses and Nurse Leadership and Management. We describe the teamwork needed and steps taken during this process, and demonstrate the adaptability, usability, and value of this tool for other online development projects. The EU-DEF is an evidence-based, comprehensive framework that can be used as a starting point for individuals or teams when developing digital educational projects. Considering all variables helps focus the expertise required to produce robust end-products.

Diagnostic Accuracy of a Portable ECG Device in Rowing Athletes.

Background: Athletes can experience exercise-induced transient arrythmias during high-intensity exercise or competition, which are difficult to capture on traditional Holter monitors or replicate in clinical exercise testing. The aim of this study was to investigate the reliability of a portable single channel ECG sensor and data recorder (PluxECG) and to evaluate the confidence and reliability in interpretation of ECGs recorded using the PluxECG during remote rowing. Methods: This was a two-phase study on rowing athletes. Phase I assessed the accuracy and precision of heart rate (HR) using the PluxECG system compared to a reference 12-lead ECG system. Phase II evaluated the confidence and reliability in interpretation of ECGs during ergometer (ERG) and on-water (OW) rowing at moderate and high intensities. ECGs were reviewed by two expert readers for HR, rhythm, artifact and confidence in interpretation. Results: Findings from Phase I found that 91.9% of samples were within the 95% confidence interval for the instantaneous value of the changing exercising HR. The mean correlation coefficient across participants and tests was 0.9886 (σ = 0.0002, SD = 0.017) and between the two systems at elevated HR was 0.9676 (σ = 0.002, SD = 0.05). Findings from Phase II found significant differences for the presence of artifacts and confidence in interpretation in ECGs between readers’ for both intensities and testing conditions. Interpretation of ECGs for OW rowing had a lower level of reader agreement than ERG rowing for HR, rhythm, and artifact. Using consensus data between readers’ significant differences were apparent between OW and ERG rowing at high-intensity rowing for HR (p = 0.05) and artifact (p = 0.01). ECGs were deemed of moderate-low quality based on confidence in interpretation and the presence of artifacts. Conclusions: The PluxECG device records accurate and reliable HR but not ECG data during exercise in rowers. The quality of ECG tracing derived from the PluxECG device is moderate-low, therefore the confidence in ECG interpretation using the PluxECG device when recorded on open water is inadequate at this time.

Atrioventricular Synchronization for Detection of Atrial Fibrillation and Flutter in One to Twelve ECG Leads Using a Dense Neural Network Classifier.

This study investigates the use of atrioventricular (AV) synchronization as an important diagnostic criterion for atrial fibrillation and flutter (AF) using one to twelve ECG leads. Heart rate, lead-specific AV conduction time, and P-/f-wave amplitude were evaluated by three representative ECG metrics (mean value, standard deviation), namely RR-interval (RRi-mean, RRi-std), PQ-interval (PQi-mean, PQI-std), and PQ-amplitude (PQa-mean, PQa-std), in 71,545 standard 12-lead ECG records from the six largest PhysioNet CinC Challenge 2021 databases. Two rhythm classes were considered (AF, non-AF), randomly assigning records into training (70%), validation (20%), and test (10%) datasets. In a grid search of 19, 55, and 83 dense neural network (DenseNet) architectures and five independent training runs, we optimized models for one-lead, six-lead (chest or limb), and twelve-lead input features. Lead-set performance and SHapley Additive exPlanations (SHAP) input feature importance were evaluated on the test set. Optimal DenseNet architectures with the number of neurons in sequential [1st, 2nd, 3rd] hidden layers were assessed for sensitivity and specificity: DenseNet [16,16,0] with primary leads (I or II) had 87.9–88.3 and 90.5–91.5%; DenseNet [32,32,32] with six limb leads had 90.7 and 94.2%; DenseNet [32,32,4] with six chest leads had 92.1 and 93.2%; and DenseNet [128,8,8] with all 12 leads had 91.8 and 95.8%, indicating sensitivity and specificity values, respectively. Mean SHAP values on the entire test set highlighted the importance of RRi-mean (100%), RR-std (84%), and atrial synchronization (40–60%) for the PQa-mean (aVR, I), PQi-std (V2, aVF, II), and PQi-mean (aVL, aVR). Our focus on finding the strongest AV synchronization predictors of AF in 12-lead ECGs would lead to a comprehensive understanding of the decision-making process in advanced neural network classifiers. DenseNet self-learned to rely on a few ECG behavioral characteristics: first, characteristics usually associated with AF conduction such as rapid heart rate, enhanced heart rate variability, and large PQ-interval deviation in V2 and inferior leads (aVF, II); second, characteristics related to a typical P-wave pattern in sinus rhythm, which is best distinguished from AF by the earliest negative P-peak deflection of the right atrium in the lead (aVR) and late positive left atrial deflection in lateral leads (I, aVL). Our results on lead-selection and feature-selection practices for AF detection should be considered for one- to twelve-lead ECG signal processing settings, particularly those measuring heart rate, AV conduction times, and P-/f-wave amplitudes. Performances are limited to the AF diagnostic potential of these three metrics. SHAP value importance can be used in combination with a human expert’s ECG interpretation to change the focus from a broad observation of 12-lead ECG morphology to focusing on the few AV synchronization findings strongly predictive of AF or non-AF arrhythmias. Our results are representative of AV synchronization findings across a broad taxonomy of cardiac arrhythmias in large 12-lead ECG databases.

Lightweight Multireceptive Field CNN for 12-Lead ECG Signal Classification.

The electrical activity produced during the heartbeat is measured and recorded by an ECG. Cardiologists can interpret the ECG machine's signals and determine the heart's health condition and related causes of ECG signal abnormalities. However, cardiologist shortage is a challenge in both developing and developed countries. Moreover, the experience of a cardiologist matters in the accurate interpretation of the ECG signal, as the interpretation of ECG is quite tricky even for experienced doctors. Therefore, developing computer-aided ECG interpretation is required for its wide-reaching effect. 12-lead ECG generates a 1D signal with 12 channels among the well-known time-series data. Classical machine learning can develop automatic detection, but deep learning is more effective in the classification task. 1D-CNN is being widely used for CVDS detection from ECG datasets. However, adopting a deep learning model designed for computer vision can be problematic because of its massive parameters and the need for many samples to train. In many detection tasks ranging from semantic segmentation of medical images to time-series data classification, multireceptive field CNN has improved performance. Notably, the nature of the ECG dataset made performance improvement possible by using a multireceptive field CNN (MRF-CNN). Using MRF-CNN, it is possible to design a model that considers semantic context information within ECG signals with different sizes. As a result, this study has designed a multireceptive field CNN architecture for ECG classification. The proposed multireceptive field CNN architecture can improve the performance of ECG signal classification. We have achieved a 0.72 F1 score and 0.93 AUC for 5 superclasses, a 0.46 F1 score and 0.92 AUC for 20 subclasses, and a 0.31 F1 score and 0.92 AUC for all the diagnostic classes of the PTB-XL dataset.

Scientific and technical Development of transtelephone Electrocardiography in foreign Countries during the 60-80s of the twentieth Century

In the context of the scientific and technical history of electrocardiography, the development of transtelephone electrocardiography (tele-ECG) remains unexplored - as a set of technical and methodological aspects of remote ECG transmission via telecommunication. This direction played a significant role in medical science and practice in the middle of the twentieth century, in many ways it remains relevant to this day. The task of the study. To identify and systematize the patterns of development of scientific knowledge related to the creation and use of transtelephone electrocardiography technologies in the period of the 1960s-1980s in Europe, Asia, North America and Australia. Results. From the point of view of technical sciences, there were two main trends in scientific research: a) the design of hardware solutions for broadcasting and receiving ECG via cable or radio channels; b) the development of algorithms for computer analysis of ECG. From the point of view of medical sciences: a) clinical trials, including an assessment of the technical reliability of equipment; b) evaluation of the diagnostic accuracy of remote (both "human" and machine) interpretation of ECG; c) study of the clinical significance of tele-ECG. When comparing the trends of scientific and technological development of transtelephone electrocardiography in foreign countries, it is obvious that the countries of Europe and Asia follow in the wake of medical science in the USA. In the studies of scientists from Eastern Europe, the influence of medical science of the USSR in the form of the use of tele-ECG in occupational medicine is noted. Conclusions. In the period of the 1960s-1980s, the scientific and technical development of transtelephone electrocardiography took place in many countries of the world, but the activity and significance of these processes seriously differed. The emergence of affordable and reliable technologies for remote counseling, as well as automated analysis of electrocardiography results, has become significant. The overall result was the emergence of clinically significant remote diagnostics techniques and new models of medical care organization based on them.

Session 9: Selected abstracts 2Various reduced lead settings for diagnostic ECG interpretation by deep learning models with a large training set and across hospital testing set

Session 9: Selected abstracts 2Various reduced lead settings for diagnostic ECG interpretation by deep learning models with a large training set and across hospital testing set