Artificial Intelligence Interpretation of the Electrocardiogram: A State-of-the-Art Review.

PURPOSE OF REVIEW: Artificial intelligence (AI) is transforming electrocardiography (ECG) interpretation. AI diagnostics can reach beyond human capabilities, facilitate automated access to nuanced ECG interpretation, and expand the scope of cardiovascular screening in the population. AI can be applied to the standard 12-lead resting ECG and single-lead ECGs in external monitors, implantable devices, and direct-to-consumer smart devices. We summarize the current state of the literature on AI-ECG. RECENT FINDINGS: Rhythm classification was the first application of AI-ECG. Subsequently, AI-ECG models have been developed for screening structural heart disease including hypertrophic cardiomyopathy, cardiac amyloidosis, aortic stenosis, pulmonary hypertension, and left ventricular systolic dysfunction. Further, AI models can predict future events like development of systolic heart failure and atrial fibrillation. AI-ECG exhibits potential in acute cardiac events and non-cardiac applications, including acute pulmonary embolism, electrolyte abnormalities, monitoring drugs therapy, sleep apnea, and predicting all-cause mortality. Many AI models in the domain of cardiac monitors and smart watches have received Food and Drug Administration (FDA) clearance for rhythm classification, while others for identification of cardiac amyloidosis, pulmonary hypertension and left ventricular dysfunction have received breakthrough device designation. As AI-ECG models continue to be developed, in addition to regulatory oversight and monetization challenges, thoughtful clinical implementation to streamline workflows, avoiding information overload and overwhelming of healthcare systems with false positive results is necessary. Research to demonstrate and validate improvement in healthcare efficiency and improved patient outcomes would be required before widespread adoption of any AI-ECG model.

Pilot study to evaluate left-to-right ventricular offset in biventricular pacing-comparison of electrocardiographic imaging and ECG.

INTRODUCTION: Biventricular pacing (BiVp) improves outcomes in systolic heart failure patients with electrical dyssynchrony. BiVp is delivered from epicardial left ventricular (LV) and endocardial right ventricular (RV) electrodes. Acute electrical activation changes with different LV-RV stimulation offsets can help guide individually optimized BiVp programming. We sought to study the BiVp ventricular activation with different LV-RV offsets and compare with 12-lead ECG. METHODS: In five patients with BiVp (63 ± 17-year-old, 80% male, LV ejection fraction 27 ± 6%), we evaluated acute ventricular epicardial activation, varying LV-RV offsets in 20 ms increments from -40 to 80 ms, using electrocardiographic imaging (ECGI) to obtain absolute ventricular electrical uncoupling (VEUabs, absolute difference in average LV and average RV activation time) and total activation time (TAT). For each patient, we calculated the correlation between ECGI and corresponding ECG (3D-QRS-area and QRS duration) with different LV-RV offsets. RESULTS: The LV-RV offset to attain minimum VEUabs in individual patients ranged 20-60 ms. In all patients, a larger LV-RV offset was required to achieve minimum VEUabs (36 ± 17 ms) or 3D-QRS-area (40 ± 14 ms) than that for minimum TAT (-4 ± 9 ms) or QRS duration (-8 ± 11 ms). In individual patients, 3D-QRS-area correlated with VEUabs (r 0.65 ± 0.24) and QRS duration correlated with TAT (r 0.95 ± 0.02). Minimum VEUabs and minimum 3D-QRS-area were obtained by LV-RV offset within 20 ms of each other in all five patients. CONCLUSIONS: LV-RV electrical uncoupling, as assessed by ECGI, can be minimized by optimizing LV-RV stimulation offset. 3D-QRS-area is a surrogate to identify LV-RV offset that minimizes LV-RV uncoupling.

Safety of catheter ablation in patients with recently implanted cardiac implantable electronic device: A 5-year experience.

INTRODUCTION: Catheter ablation (CA) can interfere with cardiac implantable electronic device (CIED) function. The safety of CA in the 1st year after CIED implantation/lead revision is uncertain. METHODS: This single center, retrospective cohort included patients who underwent CA between 2012 and 2017 and had a CIED implant/lead revision within the preceding year. We assessed the frequency of device/lead malfunctions in this population. RESULTS: We identified 1810 CAs in patients between 2012 and 2017, with 170 CAs in 163 patients within a year of a CIED implant/lead revision. Mean age 68 ± 12 years (68% men). Time between the CIED procedure and CA was 158 ± 99 days. The CA procedures included AF ablation (n = 57, 34%), AV node ablation (n = 40, 24%), SVT ablation (n = 37, 22%), and PVC/VT ablations (n = 36, 21%). The cumulative frequency of lead dislodgement, significant CIED dysfunction, and/or CIED-related infection following CA was (n = 1/170, 0.6%). There was a single atrial lead dislodgement (0.6%). There were no instances of power-on-reset or CIED-related infection. Following CA, there was no significant difference in RA or RV lead sensing (p = 0.52 and 0.84 respectively) or thresholds (p = 0.94 and 0.17 respectively). The RA impedance slightly decreased post-CA from 474 ± 80 Ohms to 460 ± 73 Ohms (p = 0.002), as did the RV impedance (from 515 ± 111 Ohms to 497 ± 98 Ohms, p < 0.0001). CONCLUSIONS: CA can be performed within 1 year following CIED implantation/lead revision with a low risk of CIED/lead malfunction or lead dislodgement. The ideal time to perform CA after a CIED remains uncertain.

Different methods of 3D QRS area calculation from vectorcardiographic X, Y, and Z Leads.

3DQRSarea is a strong marker for cardiac resynchronization therapy and can be obtained by taking the (i) summation or the (ii) difference of the areas subtended by positive and negative deflections in X, Y, Z vectorcardiographic electrocardiogram (ECG) leads. We correlated both methods with the instantaneous-absolute-3D-voltage-time-integral (VTIQRS-3D). 3DQRSarea consistently underestimated the VTIQRS -3D, but the summation method was a closer and more reliable approximation. The dissimilarity was less apparent in left bundle branch block (r2 summation .996 vs. difference .972) and biventricular paced ECGs (r2 .996 vs. .957) but was more apparent in normal ECGs (r2 .988 vs. .653).

Cardiac resynchronization therapy for pacing induced cardiomyopathy: Role of baseline right ventricular pacing burden.

BACKGROUND: Cardiac resynchronization therapy (CRT) is indicated for patients with heart failure with reduced left ventricular ejection fraction (LVEF) and chronic right ventricular (RV) pacing burden ≥40% (pacing-induced cardiomyopathy, PICM). It is uncertain whether baseline RV pacing burden impacts response to CRT. METHODS: We conducted a retrospective study of all CRT upgrades for PICM at our hospital from January 2017 to December 2018. Univariate and multivariable-adjusted changes in LVEF, and echocardiographic response (≥10% improvement in LVEF) at 3-12 months post-CRT upgrade were compared in those with RV pacing burden ≥90% versus <90%. RESULTS: We included 75 patients (age 74 ± 11 years, 71% male) who underwent CRT upgrade for PICM. The baseline RV pacing burden was ≥90% in 56 patients (median 99% [IQR 98%-99%]), and <90% in 19 patients (median 79% [IQR 73%-87%]). Improvement in LVEF was greater in those with baseline RV pacing burden ≥90% versus <90% (15.7 ± 9.3% vs. 7.5 ± 9.6%, p = .003). Baseline RV pacing burden ≥90% was a strong predictor of an improvement in LVEF ≥10% after CRT upgrade both in univariate and multivariate-adjusted models (p = .005 and .02, respectively). CONCLUSION: A higher baseline RV pacing burden predicts a greater improvement in LVEF after CRT upgrade for PICM.

Electrocardiographic Z-axis QRS-T voltage-time-integral in patients with typical right bundle branch block - Correlation with echocardiographic right ventricular size and function.

BACKGROUND: Right bundle branch block (RBBB) can be benign or associated with right ventricular (RV) functional and structural abnormalities. Our aim was to evaluate QRS-T voltage-time-integral (VTI) compared to QRS duration and lead V1 R' as markers for RV abnormalities. METHODS: We included adults with an ECG demonstrating RBBB and echocardiogram obtained within 3 months of each other, between 2010 and 2020. VTIQRS and VTIQRST were obtained for 12 standard ECG leads, reconstructed vectorcardiographic X, Y, Z leads and root-mean-squared (3D) ECG. Age, sex and BSA-adjusted linear regressions were used to assess associations of QRS duration, amplitudes, VTIs and lead V1 R' duration/VTI with echocardiographic tricuspid annular plane systolic excursion (TAPSE), RV tissue Doppler imaging S', basal and mid diameter, and systolic pressure (RVSP). RESULTS: Among 782 patients (33% women, age 71 ± 14 years) with RBBB, R' duration in lead V1 was modestly associated with RV S', RV diameters and RVSP (all p ≤ 0.03). QRS duration was more strongly associated with RV diameters (both p < 0.0001). AmplitudeQRS-Z was modestly correlated with all 5 RV echocardiographic variables (all p ≤ 0.02). VTIR'-V1 was more strongly associated with TAPSE, RV S' and RVSP (all p ≤ 0.0003). VTIQRS-Z and VTIQRST-Z were among the strongest correlates of the 5 RV variables (all p < 0.0001). VTIQRST-Z.√BSA cutoff of ≥62 µVsm had sensitivity 62.7% and specificity 65.7% for predicting ≥3 of 5 abnormal RV variables (AUC 0.66; men 0.71, women 0.60). CONCLUSION: In patients with RBBB, VTIQRST-Z is a stronger predictor of RV dysfunction and adverse remodeling than QRS duration and lead V1 R'.

Case of a misleading ECG.

An elderly man with severe chronic obstructive pulmonary disease and a history of complete heart block with pacemaker placement was found to have pacemaker lead infection and required device extraction. He had a standard dual chamber pacemaker in place, however the ECG obtained showed paced QRS complexes with presence of R wave in lead V1 and QS in lead I suggestive of left ventricular pacing. Additional imaging with CT scan obtained for confirmation revealed that the heart was displaced to the left posterior hemithorax secondary to pulmonary disease. Due to significant posterolateral rotation of the heart, a right ventricular paced rhythm can demonstrate Q/S waves in the lateral leads (I, aVL, V5-6) and R waves in the right precordial leads (V1-3). This can be misdiagnosed as a left ventricular paced rhythm.

Novel mutation in N-terminal fragment of ryanodine receptor 2 causing catecholaminergic polymorphic ventricular tachycardia.

CPVT is a rare inherited arrhythmogenic disorder characterized by bidirectional, polymorphic ventricular arrhythmias triggered by catecholamines released during exercise, stress, or sudden emotion in individuals with a normal resting electrocardiogram and structurally normal heart. Mutations in the ryanodine receptor 2 gene are the most common known etiology of this disorder. The c.1195A > G(p.Met399Val) variant in Exon 14 of RyR2 is currently classified as a Variant of Uncertain Significance. We present a case of CPVT caused by this novel disease-causing RyR2 variant and discuss its pathophysiology. The role of SSRIs in treating patients with CPVT unresponsive to mainstream therapies is also highlighted.

Electrocardiographic prediction of left ventricular hypertrophy in women and men with left bundle branch block - Comparison of QRS duration, amplitude and voltage-time-integral.

BACKGROUND: Standard ECG criteria for left ventricular (LV) hypertrophy rely on QRS amplitudes. However, in the setting of left bundle branch block (LBBB), ECG correlates of LV hypertrophy are not well established. We sought to evaluate quantitative ECG predictors of LV hypertrophy in the presence of LBBB. METHODS: We included adult patients with typical LBBB having ECG and transthoracic echocardiogram performed within 3 months of each other in 2010-2020. Orthogonal X, Y, Z leads were reconstructed from digital 12­lead ECGs using Kors's matrix. In addition to QRS duration, we evaluated QRS amplitudes and voltage-time-integrals (VTIs) from all 12 leads, X, Y, Z leads and 3D (root-mean-squared) ECG. We used age, sex and BSA-adjusted linear regressions to predict echocardiographic LV calculations (mass, end-diastolic and end-systolic volumes, ejection fraction) from ECG, and separately generated ROC curves for predicting echocardiographic abnormalities. RESULTS: We included 413 patients (53% women, age 73 ± 12 years). All 4 echocardiographic LV calculations were most strongly correlated with QRS duration (all p < 0.00001). In women, QRS duration ≥ 150 ms had sensitivity/specificity 56.3%/64.4% for increased LV mass and 62.7%/67.8% for increased LV end-diastolic volume. In men, QRS duration ≥ 160 ms had a sensitivity/specificity 63.1%/72.1% for increased LV mass and 58.3%/74.5% for increased LV end-diastolic volume. QRS duration was best able to discriminate eccentric hypertrophy (area under ROC curve 0.701) and increased LV end-diastolic volume (0.681). CONCLUSIONS: In patients with LBBB, QRS duration (≥ 150 in women and ≥ 160 in men) is a superior predictor of LV remodeling esp. eccentric hypertrophy and dilation.

Multicenter assessment of the outcomes of subcutaneous ICD implantation in patients with prior or future sternotomy.

BACKGROUND: The subcutaneous ICD (S-ICD) is a viable alternative to transvenous ICD and avoids intravascular complications in patients without a pacing indication. The outcomes of S-ICD implantation are uncertain in patients with prior sternotomy. OBJECTIVE: We aim to compare the implant techniques and outcomes with S-ICD implantation in patients with and without prior sternotomy. METHODS: Multicenter retrospective cohort study including adult patients with an S-ICD implanted between January 2014 and June 2020. Outcomes were compared between patients with and without prior sternotomy. RESULTS: Among the 212 patients (49 ± 15 years old, 43% women, BMI 30 ± 8 kg/m2 , 68% primary prevention, 30% ischemic cardiomyopathy, LVEF median 30% IQR 25%-45%) who underwent S-ICD implantation, 47 (22%) had a prior sternotomy. There was no difference in the sensing vector (57% vs. 53% primary, p = 0.55), laterality of the S-ICD lead to the sternum (94% vs. 96% leftward, p = 0.54), or the defibrillation threshold (65 ± 1.4 J vs. 65 ± 0.8 J, p = 0.76) with versus without prior sternotomy. The frequency of 30-day complications was similar with and without prior sternotomy (n = 3/47 vs. n = 15/165, 6% vs. 9%, p = 0.56). Over a median follow-up of 28 months (IQR 10-49 months), the frequency of inappropriate shocks was similar between those with and without prior sternotomy (n = 3/47 and n = 16/165, 6% vs. 10%, p = 0.58). CONCLUSION: Implantation of an S-ICD in patients with prior sternotomy is safe with a similar risk of 30-day complications and inappropriate ICD shocks as patients without prior sternotomy.

Narrow Complex Tachycardia With Intermittent 2:1 Atrioventricular Relationship.

Narrow QRS complex tachycardia has a broad differential diagnosis. We present a series of rhythm strips with representative onset, transition from 2:1 AV conduction to 1:1 AV conduction, and offset of tachycardia. By analyzing these rhythm strips, we can identify the electrophysiologic mechanism and diagnosis. (Level of Difficulty: Intermediate.).

UltraSound Axillary Vein Access (USAA): Learning curve and randomized comparison to traditional venous access for cardiac device implantation.

BACKGROUND: Many techniques exist for venous access (VA) during cardiac implantable electronic device (CIED) implantation. OBJECTIVE: We sought to evaluate the learning curve with ultrasound (US) guided axillary vein access (USAA). METHODS: Single-center prospective randomized controlled trial of patients undergoing CIED implantation. Patients were randomized in a 2:1 fashion to USAA versus conventional VA techniques. The primary outcomes were the success rates, VA times and 30-day complication rates. RESULTS: The study included 100 patients (age 68 ± 14 years, BMI 27 ± 4 kg/m2 ). USAA was successful in 66/70 implants (94%). Initial attempts at conventional VA included 47% axillary (n = 14), 30% (n = 9) cephalic, and 23% (n = 7) subclavian. The median access time was longer for USAA than conventional access (8.3 IQR 4.2-15.3 min vs. 5.2 IQR 3.4-8.6 min, p = .009). Among the five inexperienced USAA implanters, there was a significant improvement in median access time from first to last tertile of USAA implants (17.0 IQR 7.0-21.0 min to 8.6 IQR 4.5-10.8 min, p = .038). The experienced USAA implanter had similar access times with USAA compared with conventional access (4.0 IQR 3.3-4.7 min vs. 5.2 IQR 3.4-8.6 min, p = .15). Venograms were less common with USAA than conventional access (2% vs. 33%, p < .0001). The 30-day complication rate was similar with USAA (n = 4/70, 6%) versus conventional (n = 3/30, 10%, p = .44). CONCLUSION: Although the success rate with USAA was high, there was a significant learning curve. Once experienced with the USAA technique, there is the potential for reduced complications without adding to the procedure duration.

Impact of ultra-conservative ICD programming in patients with LVADs: Avoiding potentially unnecessary tachy-therapies.

BACKGROUND: Patients with left ventricular assist devices (LVAD) often tolerate ventricular arrhythmias (VA). We aim to assess the frequency and outcomes of ICD therapies averted by ultraconservative ICD programming (UCP) in LVAD patients. METHODS: This single center, retrospective cohort study included patients with LVADs and ICDs implanted from 2015 to 2019 that had UCP. The aim for UCP was to maximally delay VA treatments and maximize anti-tachycardia pacing (ATP) prior to ICD shocks. VA events were reviewed after UCP and evaluated under prior conservative programming to assess for potentially averted events (that would have resulted in either ATP or defibrillation with prior programming). RESULTS: Fifty patients were included in the study with follow-up of median 16 ± 10.2 months after UCP. The median time from LVAD implantation to reprogramming was 7 days (IQR 5-9 days). Fourteen patients (28%) had potentially averted VA events that would have been treated with their prior ICD programming (82 total events, median two events per patient, IQR 1-10 events). Treated VA events occurred in 15 patients (30%). Eleven of the 14 patients with potentially averted VAs had treated events as well. Only one patient reported definitive symptoms of self-limited "dizziness" during a potentially averted event that did not result in hospitalization. No patients died of complications from or needed emergent care/hospitalization due a potentially averted VA. CONCLUSIONS: UCP in LVAD patients likely prevented unnecessary VA treatments in many patients with minimal reported symptoms during these potentially averted events. Prospective studies are necessary to confirm these findings.

Electromagnetic interference from left ventricular assist devices detected in patients with implantable cardioverter-defibrillators.

INTRODUCTION: Electromagnetic interference (EMI) from left ventricular assist devices (LVADs) can cause implantable cardioverter-defibrillator (ICD) oversensing. We sought to assess the frequency of inappropriate shocks/oversensing due to LVAD-related EMI and prospectively compare integrated (IB) versus dedicated bipolar (DB) sensing in patients with LVADs. METHODS: Single-center study in LVAD patients with Medtronic or Abbott ICDs between September 2017 and March 2020. We excluded patients that were pacemaker dependent. Measurements were obtained of IB and DB sensing and noise to calculate a signal-to-noise ratio (SNR). Device checks were reviewed to assess appropriate and inappropriate sensing events. RESULTS: Forty patients (age 52 ± 14 years, 75% men, 38% ischemic cardiomyopathy) were included with the median time between LVAD implantation and enrollment of 6.7 months (2.3, 11.4 months). LVAD subtypes included: HeartWare (n = 22, 55%), Heartmate II (n = 10, 25%), and Heartmate III (n = 8, 20%). Over a follow-up duration of 21.6 ± 12.9 months after LVAD implantation, 5% of patients (n = 2) had oversensing of EMI from the LVAD (both with HeartWare LVADs and Abbott ICDs) at 4 days and 10.8 months after LVAD implantation. Both patients underwent adjustment of ventricular sensing with resolution of oversensing and no further events over 5 and 15 months of further follow-up. The SNR was similar between IB and DB sensing (50 [29-67] and 57 [41-69], p = 0.89). CONCLUSION: ICD oversensing of EMI from LVADs is infrequent and can be managed with reprogramming the sensitivity. There was no significant difference in the R-wave SNR with IB versus DB ICD leads.

Rapid Exclusion of COVID Infection With the Artificial Intelligence Electrocardiogram.

OBJECTIVE: To rapidly exclude severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection using artificial intelligence applied to the electrocardiogram (ECG). METHODS: A global, volunteer consortium from 4 continents identified patients with ECGs obtained around the time of polymerase chain reaction-confirmed COVID-19 diagnosis and age- and sex-matched controls from the same sites. Clinical characteristics, polymerase chain reaction results, and raw electrocardiographic data were collected. A convolutional neural network was trained using 26,153 ECGs (33.2% COVID positive), validated with 3826 ECGs (33.3% positive), and tested on 7870 ECGs not included in other sets (32.7% positive). Performance under different prevalence values was tested by adding control ECGs from a single high-volume site. RESULTS: The area under the curve for detection of acute COVID-19 infection in the test group was 0.767 (95% CI, 0.756 to 0.778; sensitivity, 98%; specificity, 10%; positive predictive value, 37%; negative predictive value, 91%). To more accurately reflect a real-world population, 50,905 normal controls were added to adjust the COVID prevalence to approximately 5% (2657/58,555), resulting in an area under the curve of 0.780 (95% CI, 0.771 to 0.790) with a specificity of 12.1% and a negative predictive value of 99.2%. CONCLUSION: Infection with SARS-CoV-2 results in electrocardiographic changes that permit the artificial intelligence-enhanced ECG to be used as a rapid screening test with a high negative predictive value (99.2%). This may permit the development of electrocardiography-based tools to rapidly screen individuals for pandemic control.

Incidence, Predictors, and Outcomes of In-Hospital Cardiac Arrest in COVID-19 Patients Admitted to Intensive and Non-Intensive Care Units: Insights From the AHA COVID-19 CVD Registry.

Background Limited information is available regarding in-hospital cardiac arrest (IHCA) in patients with COVID-19. Methods and Results We leveraged the American Heart Association COVID-19 Cardiovascular Disease (AHA COVID-19 CVD) Registry to conduct a cohort study of adults hospitalized for COVID-19. IHCA was defined as those with documentation of cardiac arrest requiring medication or electrical shock for resuscitation. Mixed effects models with random intercepts were used to identify independent predictors of IHCA and mortality while accounting for clustering at the hospital level. The study cohort included 8518 patients (6080 not in the intensive care unit [ICU]) with mean age of 61.5 years (SD 17.5). IHCA occurred in 509 (5.9%) patients overall with 375 (73.7%) in the ICU and 134 (26.3%) patients not in the ICU. The majority of patients at the time of ICHA were not in a shockable rhythm (76.5%). Independent predictors of IHCA included older age, Hispanic ethnicity (odds ratio [OR], 1.9; CI, 1.4-2.4; P<0.001), and non-Hispanic Black race (OR, 1.5; CI, 1.1-1.9; P=0.004). Other predictors included oxygen use on admission, quick Sequential Organ Failure Assessment score on admission, and hypertension. Overall, 35 (6.9%) patients with IHCA survived to discharge, with 9.1% for ICU and 0.7% for non-ICU patients. Conclusions Older age, Black race, and Hispanic ethnicity are independent predictors of IHCA in patients with COVID-19. Although the incidence is much lower than in ICU patients, approximately one-quarter of IHCA events in patients with COVID-19 occur in non-ICU settings, with the latter having a substantially lower survival to discharge rate.

His bundle lead implantation for leadless pacemaker pacing-induced cardiomyopathy in persistent left superior vena cava.

Persistent left superior vena cava (PLSVC) poses technical challenges to implantation of transvenous cardiac implantable electronic devices. His-bundle pacing is a physiologic pacing strategy to avoid or treat pacing-induced cardiomyopathy. We report a case of His-bundle lead implantation in a patient with PLSVC, absent right SVC, and pacing-induced cardiomyopathy.