Prediction of Sudden Cardiac Death Manifesting With Documented Ventricular Fibrillation or Pulseless Ventricular Tachycardia

Predicting long-term risk of sudden cardiac death with automatic computer-interpretations of electrocardiogram

Background

Computer-interpreted electrocardiogram (CIE) data is provided by almost all commercial software used to capture and store digital electrocardiograms. CIE is widely available, inexpensive, and accurate. We tested the potential of CIE in long-term sudden cardiac death (SCD) risk prediction.

Methods

This is a retrospective of 8,568 consecutive patients treated for acute coronary syndrome. The primary endpoint was five-year occurrence of SCDs or equivalent events (SCDs aborted by successful resuscitation or adequate ICD therapy). CIE statements were extracted from summary statements and measurements made by the GE Muse 12SL algorithm from ECGs taken during admission. Three supervised machine learning algorithms (logistic regression, extreme gradient boosting, and random forest) were then used for analysis to find risk features using a random 70/30% split for discovery and validation cohorts.

Results

Five-year SCD occurrence rate was 3.3% (n = 287). Regardless of the used ML algorithm, the most significant risk ECG risk features detected by the CIE included known risk features such as QRS duration and factors associated with QRS duration, heart rate-corrected QT time (QTc), and the presence of premature ventricular contractions (PVCs). Risk score formed by using most significant CIE features associated with the risk of SCD despite adjusting for any clinical risk factor (including left ventricular ejection fraction). Sensitivity of CIE data to correctly identify patients with high risk of SCD (over 10% 5-year risk of SCD) was usually low, but specificity and negative prediction value reached up to 96.9% and 97.3% when selecting only the most significant features identified by logistic regression modeling (p-value threshold <0.01 for accepting features in the model). Overall, CIE data showed a modest overall performance for identifying high risk individuals with area under the receiver operating characteristic curve values ranging between 0.652 and 0.693 (highest for extreme gradient boosting and lowest for logistic regression).

Conclusion

This proof-of-concept study shows that automatic interpretation of ECG identifies previously validated risk features for SCD.

Prediction of sudden cardiac death using artificial intelligence: Current status and future directions

Sudden cardiac death (SCD) remains a pressing health issue, affecting hundreds of thousands each year globally. The heterogeneity among people who suffer a SCD, ranging from individuals with severe heart failure to seemingly healthy individuals, poses a significant challenge for effective risk assessment. Conventional risk stratification, which primarily relies on left ventricular ejection fraction, has resulted in only modest efficacy of implantable cardioverter-defibrillators for SCD prevention. In response, artificial intelligence (AI) holds promise for personalized SCD risk prediction and tailoring preventive strategies to the unique profiles of individual patients. Machine and deep learning algorithms have the capability to learn intricate nonlinear patterns between complex data and defined end points, and leverage these to identify subtle indicators and predictors of SCD that may not be apparent through traditional statistical analysis. However, despite the potential of AI to improve SCD risk stratification, there are important limitations that need to be addressed. We aim to provide an overview of the current state-of-the-art of AI prediction models for SCD, highlight the opportunities for these models in clinical practice, and identify the key challenges hindering widespread adoption.

Predicting Out-of-Hospital Cardiac Arrest in the General Population Using Electronic Health Records

BACKGROUND

The majority of out-of-hospital cardiac arrests (OHCAs) occur among individuals in the general population, for whom there is no established strategy to identify risk. In this study, we assess the use of electronic health record (EHR) data to identify OHCA in the general population and define salient factors contributing to OHCA risk.

METHODS

The analytical cohort included 2366 individuals with OHCA and 23 660 age- and sex-matched controls receiving health care at the University of Washington. Comorbidities, electrocardiographic measures, vital signs, and medication prescription were abstracted from the EHR. The primary outcome was OHCA. Secondary outcomes included shockable and nonshockable OHCA. Model performance including area under the receiver operating characteristic curve and positive predictive value were assessed and adjusted for observed rate of OHCA across the health system.

RESULTS

There were significant differences in demographic characteristics, vital signs, electrocardiographic measures, comorbidities, and medication distribution between individuals with OHCA and controls. In external validation, discrimination in machine learning models (area under the receiver operating characteristic curve 0.80-0.85) was superior to a baseline model with conventional cardiovascular risk factors (area under the receiver operating characteristic curve 0.66). At a specificity threshold of 99%, correcting for baseline OHCA incidence across the health system, positive predictive value was 2.5% to 3.1% in machine learning models compared with 0.8% for the baseline model. Longer corrected QT interval, substance abuse disorder, fluid and electrolyte disorder, alcohol abuse, and higher heart rate were identified as salient predictors of OHCA risk across all machine learning models. Established cardiovascular risk factors retained predictive importance for shockable OHCA, but demographic characteristics (minority race, single marital status) and noncardiovascular comorbidities (substance abuse disorder) also contributed to risk prediction. For nonshockable OHCA, a range of salient predictors, including comorbidities, habits, vital signs, demographic characteristics, and electrocardiographic measures, were identified.

CONCLUSIONS

In a population-based case-control study, machine learning models incorporating readily available EHR data showed reasonable discrimination and risk enrichment for OHCA in the general population. Salient factors associated with OCHA risk were myriad across the cardiovascular and noncardiovascular spectrum. Public health and tailored strategies for OHCA prediction and prevention will require incorporation of this complexity.

Emergence of Artificial Intelligence and Machine Learning Models in Sudden Cardiac Arrest: A Comprehensive Review of Predictive Performance and Clinical Decision Support

Sudden cardiac death/sudden cardiac arrest (SCD/SCA) is an increasingly prevalent cause of mortality globally, particularly in individuals with preexisting cardiac conditions. The ambiguous premortem warnings and the restricted interventional window related to SCD account for the complexity of the condition. Current reports suggest SCD to be accountable for 20% of all deaths hence accurately predicting SCD risk is an imminent concern. Traditional approaches for predicting SCA, particularly "track-and-trigger" warning systems have demonstrated considerable inadequacies, including low sensitivity, false alarms, decreased diagnostic liability, reliance on clinician involvement, and human errors. Artificial intelligence (AI) and machine learning (ML) models have demonstrated near-perfect accuracy in predicting SCA risk, allowing clinicians to intervene timely. Given the constraints of current diagnostics, exploring the benefits of AI and ML models in enhancing outcomes for SCA/SCD is imperative. This review article aims to investigate the efficacy of AI and ML models in predicting and managing SCD, particularly targeting accuracy in prediction.

Lack of Prognostic Value of T-Wave Alternans for Implantable Cardioverter-Defibrillator Benefit in Primary Prevention

BACKGROUND

New methods to identify patients who benefit from a primary prophylactic implantable cardioverter-defibrillator (ICD) are needed. T-wave alternans (TWA) has been shown to associate with arrhythmogenesis of the heart and sudden cardiac death. We hypothesized that TWA might be associated with benefit from ICD implantation in primary prevention.

METHODS AND RESULTS

In the EU-CERT-ICD (European Comparative Effectiveness Research to Assess the Use of Primary Prophylactic Implantable Cardioverter-Defibrillators) study, we prospectively enrolled 2327 candidates for primary prophylactic ICD. A 24-hour Holter monitor reading was taken from all recruited patients at enrollment. TWA was assessed from Holter monitoring using the modified moving average method. Study outcomes were all-cause death, appropriate shock, and survival benefit. TWA was assessed both as a contiguous variable and as a dichotomized variable with cutoff points <47 μV and <60 μV. The final cohort included 1734 valid T-wave alternans samples, 1211 patients with ICD, and 523 control patients with conservative treatment, with a mean follow-up time of 2.3 years. TWA ≥60 μV was a predicter for a higher all-cause death in patients with an ICD on the basis of a univariate Cox regression model (hazard ratio, 1.484 [95% CI, 1.024-2.151]; P=0.0374; concordance statistic, 0.51). In multivariable models, TWA was not prognostic of death or appropriate shocks in patients with an ICD. In addition, TWA was not prognostic of death in control patients. In a propensity score-adjusted Cox regression model, TWA was not a predictor of ICD benefit.

CONCLUSIONS

T-wave alternans is poorly prognostic in patients with a primary prophylactic ICD. Although it may be prognostic of life-threatening arrhythmias and sudden cardiac death in several patient populations, it does not seem to be useful in assessing benefit from ICD therapy in primary prevention among patients with an ejection fraction of ≤35%.

Dynamic electrocardiogram changes are a novel risk marker for sudden cardiac death

BACKGROUND AND AIMS

Electrocardiogram (ECG) abnormalities have been evaluated as static risk markers for sudden cardiac death (SCD), but the potential importance of dynamic ECG remodelling has not been investigated. In this study, the nature and prevalence of dynamic ECG remodelling were studied among individuals who eventually suffered SCD.

METHODS

The study population was drawn from two prospective community-based SCD studies in Oregon (2002, discovery cohort) and California, USA (2015, validation cohort). For this present sub-study, 231 discovery cases (2015-17) and 203 validation cases (2015-21) with ≥2 archived pre-SCD ECGs were ascertained and were matched to 234 discovery and 203 validation controls based on age, sex, and duration between the ECGs. Dynamic ECG remodelling was measured as progression of a previously validated cumulative six-variable ECG electrical risk score.

RESULTS

Oregon SCD cases displayed greater electrical risk score increase over time vs. controls [+1.06 (95% confidence interval +0.89 to +1.24) vs. -0.05 (-0.21 to +0.11); P < .001]. These findings were successfully replicated in California [+0.87 (+0.7 to +1.04) vs. -0.11 (-0.27 to 0.05); P < .001]. In multivariable models, abnormal dynamic ECG remodelling improved SCD prediction over baseline ECG, demographics, and clinical SCD risk factors in both Oregon [area under the receiver operating characteristic curve 0.770 (95% confidence interval 0.727-0.812) increased to area under the receiver operating characteristic curve 0.869 (95% confidence interval 0.837-0.902)] and California cohorts.

CONCLUSIONS

Dynamic ECG remodelling improved SCD risk prediction beyond clinical factors combined with the static ECG, with successful validation in a geographically distinct population. These findings introduce a novel concept of SCD dynamic risk and warrant further detailed investigation.

An ECG-based artificial intelligence model for assessment of sudden cardiac death risk

BACKGROUND

Conventional ECG-based algorithms could contribute to sudden cardiac death (SCD) risk stratification but demonstrate moderate predictive capabilities. Deep learning (DL) models use the entire digital signal and could potentially improve predictive power. We aimed to train and validate a 12 lead ECG-based DL algorithm for SCD risk assessment.

METHODS

Out-of-hospital SCD cases were prospectively ascertained in the Portland, Oregon, metro area. A total of 1,827 pre- cardiac arrest 12 lead ECGs from 1,796 SCD cases were retrospectively collected and analyzed to develop an ECG-based DL model. External validation was performed in 714 ECGs from 714 SCD cases from Ventura County, CA. Two separate control group samples were obtained from 1342 ECGs taken from 1325 individuals of which at least 50% had established coronary artery disease. The DL model was compared with a previously validated conventional 6 variable ECG risk model.

RESULTS

The DL model achieves an AUROC of 0.889 (95% CI 0.861-0.917) for the detection of SCD cases vs. controls in the internal held-out test dataset, and is successfully validated in external SCD cases with an AUROC of 0.820 (0.794-0.847). The DL model performs significantly better than the conventional ECG model that achieves an AUROC of 0.712 (0.668-0.756) in the internal and 0.743 (0.711-0.775) in the external cohort.

CONCLUSIONS

An ECG-based DL model distinguishes SCD cases from controls with improved accuracy and performs better than a conventional ECG risk model. Further detailed investigation is warranted to evaluate how the DL model could contribute to improved SCD risk stratification.

Artificial Intelligence Model Predicts Sudden Cardiac Arrest Manifesting With Pulseless Electric Activity Versus Ventricular Fibrillation

BACKGROUND

There is no specific treatment for sudden cardiac arrest (SCA) manifesting as pulseless electric activity (PEA) and survival rates are low; unlike ventricular fibrillation (VF), which is treatable by defibrillation. Development of novel treatments requires fundamental clinical studies, but access to the true initial rhythm has been a limiting factor.

METHODS

Using demographics and detailed clinical variables, we trained and tested an AI model (extreme gradient boosting) to differentiate PEA-SCA versus VF-SCA in a novel setting that provided the true initial rhythm. A subgroup of SCAs are witnessed by emergency medical services personnel, and because the response time is zero, the true SCA initial rhythm is recorded. The internal cohort consisted of 421 emergency medical services-witnessed out-of-hospital SCAs with PEA or VF as the initial rhythm in the Portland, Oregon metropolitan area. External validation was performed in 220 emergency medical services-witnessed SCAs from Ventura, CA.

RESULTS

In the internal cohort, the artificial intelligence model achieved an area under the receiver operating characteristic curve of 0.68 (95% CI, 0.61-0.76). Model performance was similar in the external cohort, achieving an area under the receiver operating characteristic curve of 0.72 (95% CI, 0.59-0.84). Anemia, older age, increased weight, and dyspnea as a warning symptom were the most important features of PEA-SCA; younger age, chest pain as a warning symptom and established coronary artery disease were important features associated with VF.

CONCLUSIONS

The artificial intelligence model identified novel features of PEA-SCA, differentiated from VF-SCA and was successfully replicated in an external cohort. These findings enhance the mechanistic understanding of PEA-SCA with potential implications for developing novel management strategies.

Dynamic prediction of malignant ventricular arrhythmias using neural networks in patients with an implantable cardioverter-defibrillator

BACKGROUND

Risk stratification for ventricular arrhythmias currently relies on static measurements that fail to adequately capture dynamic interactions between arrhythmic substrate and triggers over time. We trained and internally validated a dynamic machine learning (ML) model and neural network that extracted features from longitudinally collected electrocardiograms (ECG), and used these to predict the risk of malignant ventricular arrhythmias.

METHODS

A multicentre study in patients implanted with an implantable cardioverter-defibrillator (ICD) between 2007 and 2021 in two academic hospitals was performed. Variational autoencoders (VAEs), which combine neural networks with variational inference principles, and can learn patterns and structure in data without explicit labelling, were trained to encode the mean ECG waveforms from the limb leads into 16 variables. Supervised dynamic ML models using these latent ECG representations and clinical baseline information were trained to predict malignant ventricular arrhythmias treated by the ICD. Model performance was evaluated on a hold-out set, using time-dependent receiver operating characteristic (ROC) and calibration curves.

FINDINGS

2942 patients (61.7 ± 13.9 years, 25.5% female) were included, with a total of 32,129 ECG recordings during a mean follow-up of 43.9 ± 35.9 months. The mean time-varying area under the ROC curve for the dynamic model was 0.738 ± 0.07, compared to 0.639 ± 0.03 for a static (i.e. baseline-only model). Feature analyses indicated dynamic changes in latent ECG representations, particularly those affecting the T-wave morphology, were of highest importance for model predictions.

INTERPRETATION

Dynamic ML models and neural networks effectively leverage routinely collected longitudinal ECG recordings for personalised and updated predictions of malignant ventricular arrhythmias, outperforming static models.

FUNDING

This publication is part of the project DEEP RISK ICD (with project number 452019308) of the research programme Rubicon which is (partly) financed by the Dutch Research Council (NWO). This research is partly funded by the Amsterdam Cardiovascular Sciences (personal grant F.V.Y.T).

Sudden Cardiac Death at Home: Potential Lives Saved With Fully Automated External Defibrillators

Sudden cardiac death from ventricular arrhythmia kills about 350,000 people annually in the United States. This number has not improved since the widespread public availability of semi-automated external defibrillators (AEDs) and the teaching of nonbreathing cardiopulmonary resuscitation (CPR) procedures. When an out-of-hospital cardiac arrest occurs in a public space, lay witnesses do CPR in 40% of the cases and use AEDs on only 7.4% of the victims before emergency medical services (EMS) arrive. About 70% of sudden cardiac death occurs at home, where an AED is usually unavailable until EMS appears. The time from a 911 call to shock averages approximately 7 minutes in urban areas and is more than 14.5 minutes in rural environments. Because arrest onset is often not observed, arrest onset to shock times maybe even longer. Survival from cardiac arrest decreases by approximately 7 to 10% per minute of ventricular arrhythmia. A prearrest protocol is proposed for the at-home use of fully automated external defibrillators in select cardiac patients, which should reduce the arrest-to-shock interval to under 1 minute and may eliminate the need for CPR in some cases.

Warning symptoms associated with imminent sudden cardiac arrest: a population-based case-control study with external validation

BACKGROUND

Sudden cardiac arrest is a global public health problem with a mortality rate of more than 90%. Prearrest warning symptoms could be harnessed using digital technology to potentially improve survival outcomes. We aimed to estimate the strength of association between symptoms and imminent sudden cardiac arrest.

METHODS

We conducted a case-control study of individuals with sudden cardiac arrest and participants without sudden cardiac arrest who had similar symptoms identified from two US community-based studies of patients with sudden cardiac arrest in California state, USA (discovery population; the Ventura Prediction of Sudden Death in Multi-Ethnic Communities [PRESTO] study), and Oregon state, USA (replication population; the Oregon Sudden Unexpected Death Study [SUDS]). Participant data were obtained from emergency medical services reports for people aged 18-85 years with witnessed sudden cardiac arrest (between Feb 1, 2015, and Jan 31, 2021) and an inclusion symptom. Data were also obtained from corresponding control populations without sudden cardiac arrest who were attended by emergency medical services for similar symptoms (between Jan 1 and Dec 31, 2019). We evaluated the association of symptoms with sudden cardiac arrest in the discovery population and validated our results in the replication population by use of logistic regression models.

FINDINGS

We identified 1672 individuals with sudden cardiac arrest from the PRESTO study, of whom 411 patients (mean age 65·7 [SD 12·4] years; 125 women and 286 men) were included in the analysis for the discovery population. From a total of 76 734 calls to emergency medical services, 1171 patients (mean age 61·8 [SD 17·3] years; 643 women, 514 men, and 14 participants without data for sex) were included in the control group. Patients with sudden cardiac arrest were more likely to have dyspnoea (168 [41%] of 411 vs 262 [22%] of 1171; p<0·0001), chest pain (136 [33%] vs 296 [25%]; p=0·0022), diaphoresis (50 [12%] vs 90 [8%]; p=0·0059), and seizure-like activity (43 [11%] vs 77 [7%], p=0·011). Symptom frequencies and patterns differed significantly by sex. Among men, chest pain (odds ratio [OR] 2·2, 95% CI 1·6-3·0), dyspnoea (2·2, 1·6-3·0), and diaphoresis (1·7, 1·1-2·7) were significantly associated with sudden cardiac arrest, whereas among women, only dyspnoea was significantly associated with sudden cardiac arrest (2·9, 1·9-4·3). 427 patients with sudden cardiac arrest (mean age 62·2 [SD 13·5]; 122 women and 305 men) were included in the analysis for the replication population and 1238 patients (mean age 59·3 [16·5] years; 689 women, 548 men, and one participant missing data for sex) were included in the control group. Findings were mostly consistent in the replication population; however, notable differences included that, among men, diaphoresis was not associated with sudden cardiac arrest and chest pain was associated with sudden cardiac arrest only in the sex-stratified multivariable analysis.

INTERPRETATION

The prevalence of warning symptoms was sex-specific and differed significantly between patients with sudden cardiac arrest and controls. Warning symptoms hold promise for prediction of imminent sudden cardiac arrest but might need to be augmented with additional features to maximise predictive power.

FUNDING

US National Heart Lung and Blood Institute.

Risk Factors for Sudden Cardiac Arrest Among Hispanic or Latino Adults in Southern California: Ventura PRESTO and HCHS/SOL

Background Out-of-hospital sudden cardiac arrest (SCA) is a leading cause of mortality, making prevention of SCA a public health priority. No studies have evaluated predictors of SCA risk among Hispanic or Latino individuals in the United States. Methods and Results In this case-control study, adult SCA cases ages 18-85 (n=1,468) were ascertained in the ongoing Ventura Prediction of Sudden Death in Multi-Ethnic Communities (PRESTO) study (2015-2021) in Ventura County, California. Control subjects were selected from 3033 Hispanic or Latino participants who completed Visit 2 examinations (2014-2017) at the San Diego site of the HCHS/SOL (Hispanic Community Health Survey/Study of Latinos). We used logistic regression to evaluate the association of clinical factors with SCA. Among Hispanic or Latino SCA cases (n=295) and frequency-matched HCHS/SOL controls (n=590) (70.2% men with mean age 63.4 and 61.2 years, respectively), the following clinical variables were associated with SCA in models adjusted for age, sex, and other clinical variables: chronic kidney disease (odds ratio [OR], 7.3 [95% CI, 3.8-14.3]), heavy drinking (OR, 4.5 [95% CI, 2.3-9.0]), stroke (OR, 3.1 [95% CI, 1.2-8.0]), atrial fibrillation (OR, 3.7 [95% CI, 1.7-7.9]), coronary artery disease (OR, 2.9 [95% CI, 1.5-5.9]), heart failure (OR, 2.5 [95% CI, 1.2-5.1]), and diabetes (OR, 1.5 [95% CI, 1.0-2.3]). Conclusions In this first population-based study, to our knowledge, of SCA risk predictors among Hispanic or Latino adults, chronic kidney disease was the strongest risk factor for SCA, and established cardiovascular disease was also important. Early identification and management of chronic kidney disease may reduce SCA risk among Hispanic or Latino individuals, in addition to prevention and treatment of cardiovascular disease.

Artificial Intelligence in Ventricular Arrhythmias and Sudden Death

Sudden cardiac arrest due to lethal ventricular arrhythmias is a major cause of mortality worldwide and results in more years of potential life lost than any individual cancer. Most of these sudden cardiac arrest events occur unexpectedly in individuals who have not been identified as high-risk due to the inadequacy of current risk stratification tools. Artificial intelligence tools are increasingly being used to solve complex problems and are poised to help with this major unmet need in the field of clinical electrophysiology. By leveraging large and detailed datasets, artificial intelligence-based prediction models have the potential to enhance the risk stratification of lethal ventricular arrhythmias. This review presents a synthesis of the published literature and a discussion of future directions in this field.

Recurrent Out-of-Hospital Sudden Cardiac Arrest: Prevalence and Clinical Factors

BACKGROUND

Despite improvements in management following survival from sudden cardiac arrest (SCA) and wide availability of implantable cardioverter defibrillators for secondary prevention, a subgroup of individuals will suffer multiple distinct episodes of SCA. The objective of this study was to characterize and evaluate the burden of recurrent out-of-hospital SCA among survivors of SCA in a single large US community.

METHODS

SCA cases were prospectively ascertained in the Oregon Sudden Unexpected Death Study. Individuals that experienced recurrent SCA were identified both prospectively and retrospectively.

RESULTS

We ascertained 6649 individuals with SCA (2002-2020) and 924 (14%) survived to hospital discharge. Of these, 88 survivors (10%) experienced recurrent SCA. Of the nonsurvivors (n=5725), 35 had suffered a recurrent SCA. Of the total 123 SCA cases with recurrent SCA, >60% occurred at least 1 year after the initial SCA (median 23 months, range: 6 days to 31 years). SCA occurred despite a secondary prevention implantable cardioverter defibrillator in 22% (n=26). Prevalence of coronary disease (36% versus 25%), hypertension (69% versus 43%), diabetes (44% versus 21%), and chronic kidney disease (35% versus 14%) was significantly higher in recurrent SCA versus single SCA survivors (n=80, P=0.01). Among individuals with no secondary prevention implantable cardioverter defibrillators before recurrent SCA, the majority had apparently reversible etiologies identified at initial SCA, with one-quarter undergoing coronary revascularization and over half diagnosed with noncoronary cardiac etiologies.

CONCLUSIONS

At least 10% of SCA survivors had recurrent SCA, and a large subgroup suffered their repeat SCA despite treatment for an apparently reversible etiology. A renewed focus on careful assessment of cardiac substrate as well as management of coronary disease, hypertension, diabetes, and chronic kidney disease in SCA survivors could reduce recurrent SCA.