Continuous template collection and updating for electrogram morphology discrimination in implantable cardioverter defibrillators

Role of Machine Learning and Artificial Intelligence in Arrhythmias and Electrophysiology

Machine learning (ML), a subset of artificial intelligence (AI) centered on machines learning from extensive datasets, stands at the forefront of a technological revolution shaping various facets of society. Cardiovascular medicine has emerged as a key domain for ML applications, with considerable efforts to integrate these innovations into routine clinical practice. Within cardiac electrophysiology, ML applications, especially in the automated interpretation of electrocardiograms, have garnered substantial attention in existing literature. However, less recognized are the diverse applications of ML in cardiac electrophysiology and arrhythmias, spanning basic science research on arrhythmia mechanisms, both experimental and computational, as well as contributions to enhanced techniques for mapping cardiac electrical function and translational research related to arrhythmia management. This comprehensive review delves into various ML applications within the scope of this journal, organized into 3 parts. The first section provides a fundamental understanding of general ML principles and methodologies, serving as a foundational resource for readers interested in exploring ML applications in arrhythmia research. The second part offers an in-depth review of studies in arrhythmia and electrophysiology that leverage ML methodologies, showcasing the broad potential of ML approaches. Each subject is thoroughly outlined, accompanied by a review of notable ML research advancements. Finally, the review delves into the primary challenges and future perspectives surrounding ML-driven cardiac electrophysiology and arrhythmias research.

Machine Learning in Arrhythmia and Electrophysiology

Machine learning (ML), a branch of artificial intelligence, where machines learn from big data, is at the crest of a technological wave of change sweeping society. Cardiovascular medicine is at the forefront of many ML applications, and there is a significant effort to bring them into mainstream clinical practice. In the field of cardiac electrophysiology, ML applications have also seen a rapid growth and popularity, particularly the use of ML in the automatic interpretation of ECGs, which has been extensively covered in the literature. Much lesser known are the other aspects of ML application in cardiac electrophysiology and arrhythmias, such as those in basic science research on arrhythmia mechanisms, both experimental and computational; in the development of better techniques for mapping of cardiac electrical function; and in translational research related to arrhythmia management. In the current review, we examine comprehensively such ML applications as they match the scope of this journal. The current review is organized in 3 parts. The first provides an overview of general ML principles and methodologies that will afford readers of the necessary information on the subject, serving as the foundation for inviting further ML applications in arrhythmia research. The basic information we provide can serve as a guide on how one might design and conduct an ML study. The second part is a review of arrhythmia and electrophysiology studies in which ML has been utilized, highlighting the broad potential of ML approaches. For each subject, we outline comprehensively the general topics, while reviewing some of the research advances utilizing ML under the subject. Finally, we discuss the main challenges and the perspectives for ML-driven cardiac electrophysiology and arrhythmia research.

Primary Results From the Understanding Outcomes With the S-ICD in Primary Prevention Patients With Low Ejection Fraction (UNTOUCHED) Trial

Supplemental Digital Content is available in the text.

Background:

The subcutaneous (S) implantable cardioverter-defibrillator (ICD) is safe and effective for sudden cardiac death prevention. However, patients in previous S-ICD studies had fewer comorbidities, had less left ventricular dysfunction, and received more inappropriate shocks (IAS) than in typical transvenous ICD trials. The UNTOUCHED trial (Understanding Outcomes With the S-ICD in Primary Prevention Patients With Low Ejection Fraction) was designed to evaluate the IAS rate in a more typical, contemporary ICD patient population implanted with the S-ICD using standardized programming and enhanced discrimination algorithms.

Methods:

Primary prevention patients with left ventricular ejection fraction ≤35% and no pacing indications were included. Generation 2 or 3 S-ICD devices were implanted and programmed with rate-based therapy delivery for rates ≥250 beats per minute and morphology discrimination for rates ≥200 and <250 beats per minute. Patients were followed for 18 months. The primary end point was the IAS-free rate compared with a 91.6% performance goal, derived from the results for the ICD-only patients in the MADIT-RIT study (Multicenter Automatic Defibrillator Implantation Trial–Reduce Inappropriate Therapy). Kaplan-Meier analyses were performed to evaluate event-free rates for IAS, all-cause shock, and complications. Multivariable proportional hazard analysis was performed to determine predictors of end points.

Results:

S-ICD implant was attempted in 1116 patients, and 1111 patients were included in postimplant follow-up analysis. The cohort had a mean age of 55.8±12.4 years, 25.6% were women, 23.4% were Black, 53.5% had ischemic heart disease, 87.7% had symptomatic heart failure, and the mean left ventricular ejection fraction was 26.4±5.8%. Eighteen-month freedom from IAS was 95.9% (lower confidence limit, 94.8%). Predictors of reduced incidence of IAS were implanting the most recent generation of device, using the 3-incision technique, no history of atrial fibrillation, and ischemic cause. The 18-month all-cause shock-free rate was 90.6% (lower confidence limit, 89.0%), meeting the prespecified performance goal of 85.8%. Conversion success rate for appropriate, discrete episodes was 98.4%. Complication-free rate at 18 months was 92.7%.

Conclusions:

This study demonstrates high efficacy and safety with contemporary S-ICD devices and programming despite the relatively high incidence of comorbidities in comparison with earlier S-ICD trials. The inappropriate shock rate (3.1% at 1 year) is the lowest reported for the S-ICD and lower than many transvenous ICD studies using contemporary programming to reduce IAS.

Registration:

URL: https://www.clinicaltrials.gov; Unique identifier: NCT02433379.

Influence of myopotential interference on the Wavelet discrimination algorithm in implantable cardioverter-defibrillator

Background

Wavelet is a morphology-based algorithm for detecting ventricular tachycardia. The electrogram (EGM) source of the Wavelet algorithm is nominally programmed with the Can-RV coil configuration, which records a far-field ventricular potential. Therefore, it may be influenced by myopotential interference.

Methods

We performed a retrospective review of 40 outpatients who had an implantable cardioverter-defibrillator (ICD) with the Wavelet algorithm. The percent-match score of the Wavelet algorithm was measured during the isometric chest press by pressing the palms together. We classified patients with percent-match scores below 70% due to myopotential interference as positive morphology change, and those with 70% or more as negative morphology change. Stored episodes of tachycardia were evaluated during the follow-up.

Results

The number of patients in the positive morphology change group was 22 (55%). Amplitude of the Can-RV coil EGM was lower in the positive morphology change group compared to that in the negative group (3.9±1.3 mV vs. 7.4±1.6 mV, P=0.0015). The cut-off value of the Can-RV coil EGM was 5 mV (area under curve, 0.89). Inappropriate detections caused by myopotential interference occurred in two patients (5%) during a mean follow-up period of 49 months, and one of them received an inappropriate ICD shock. These patients had exhibited positive morphology change.

Conclusions

The Wavelet algorithm is influenced by myopotential interference when the Can-RV coil EGM is less than 5 mV.

Prospective comparison of discrimination algorithms to prevent inappropriate ICD therapy: primary results of the Rhythm ID Going Head to Head Trial

BACKGROUND

Inappropriate therapy for supraventricular arrhythmias remains a significant source of morbidity in implantable cardioverter-defibrillator (ICD) recipients.

OBJECTIVE

The Rhythm ID Goes Head to Head Trial (RIGHT) was designed to compare rhythm discrimination and inappropriate therapies among patients with ICDs from 2 manufacturers.

METHODS

Patients with standard ICD indications were randomized to receive a Guidant VITALITY 2 with Rhythm ID or selective Medtronic pulse generators using the Enhanced PR Logic or Wavelet discrimination algorithms. A single- or dual-chamber device was implanted based on clinical indications and programmed in 2 detection zones with detection enhancements enabled for rates between 150 and 200 bpm. Algorithm performance was compared between randomization groups, stratified by single or dual chamber, for the primary end point of first inappropriate therapy (shock or antitachycardia pacing) for supraventricular arrhythmias.

RESULTS

There were 1962 patients enrolled and followed for 18.3 ± 9.2 months, with no difference in all-cause mortality between groups. There were 3973 treated episodes where electrograms were available and adjudicated. The primary end point of inappropriate therapy occurred in 246 of 985 VITALITY 2 patients vs 187 of 977 specific Medtronic ICD patients (hazard ratio = 1.34; confidence interval = 1.11-1.62; P = .003). Differences in inappropriate therapy were confined to single-chamber ICDs. Inappropriate shocks were more frequent in VITALITY 2 ICDs (hazard ratio = 1.63; confidence interval = 1.29-2.06; P < .001), with most therapies and performance differences occurring at slower rhythms (rates < 175 bpm).

CONCLUSION

Rhythm discrimination performed better in the specific Medtronic than in VITALITY 2 ICDs evaluated, particularly for single-chamber devices. Inappropriate therapies, and differences in performance, may be reduced with the use of rate cutoff above 175 bpm.

Standard template of adult magnetocardiogram

BACKGROUND

We need to know the magnetocardiogram (MCG) features regarding waveform and two-dimensional current distribution in normal subjects in order to classify the abnormal waveform in patients with heart disease. However, a standard MCG waveform has not been produced yet, therefore, we have first made the standard template MCG waveform.

METHODS AND RESULTS

We used data from 464 normal control subjects' 64-channel MCGs (268 males, 196 females) to produce a template MCG waveform. The measured data were averaged after shortening or lengthening and normalization. The time interval and amplitude of the averaged data were adjusted to mean values obtained from a database. Furthermore, the current distributions (current arrow maps [CAMs]) were calculated from the produced templates to determine the current distribution pattern. The produced template of the QRS complex had a typical shape in six regions that we defined (M1, M2, M3, M4, M5, and M6). In the P wave, the main current arrow in CAMs pointing in a lower-left direction appeared in M1. In the QRS complex, the typical wave appeared in each region, and there were two main current arrows in M2 and M5. There were negative T waves in M1, M4, and M5, and positive T waves in M3 and M6, and the main current arrow pointing in a lower-left direction appeared in M2.

CONCLUSION

Template MCG waveforms were produced. These morphologic features were classified into six regions, and the current distribution was characterized in each region. Consequently, the templates and classifications enable understanding MCG features and writing clinical reports.

Tools and strategies for the reduction of inappropriate implantable cardioverter defibrillator shocks

Implantable cardioverter defibrillators (ICDs) have been shown to provide a survival benefit in patients at high risk of sudden cardiac death. A major problem associated with ICD therapy is the occurrence of inappropriate shocks which impair patients' quality of life and may also be arrhythmogenic. Despite recent technological advances, the incidence of inappropriate shocks remains high, thus posing a challenge that we have to meet. In the present review we summarise the available tools and the strategies that can be followed in order to reduce inappropriate ICD shocks.

Wavelet-Based Tachycardia Discrimination in ICDs: Impact of Posture and Electrogram Configuration

BACKGROUND

Inappropriate therapy delivery is an important concern in the management of patients with implantable cardioverter defibrillators (ICDs). Recently, a morphology-based algorithm (wavelet feature) has been introduced for differentiation of ventricular and supraventricular tachycardia. In this study, we evaluated the performance of the wavelet algorithm using various electrogram (EGM) configurations during different body positions.

METHODS

Patients with a single-chamber Medtronic model 7230 ICD (Minneapolis, MN, USA) and a double-coil lead were included. EGM templates were collected during baseline rhythm in supine position for different EGM sources (right ventricular [RV] coil-can, RV coil-superior vena cava [SVC] coil, tip-ring, SVC coil-can). For each EGM configuration, morphologic similarity (match percentage) of EGMs obtained during different body positions (supine, left and right lateral, sitting, standing, walking) were compared with the templates.

RESULTS

Twenty-eight patients (24 males; age 58 +/- 17 years) were studied. A total of 9,775 intracardiac EGMs were analyzed. Median match percentage (interquartile range) was 88% (85-94), 88% (82-94), 82% (76-88), and 73 (58-85) for the RV coil-can, RV coil-SVC coil, tip-ring, and SVC coil-can configurations, respectively. Correct classification rates, as defined by match percentage of 70% or higher, were significantly higher with the RV coil-can, RV coil-SVC coil, and tip-ring EGM configurations, as compared to the SVC coil-can configuration (95, 91, and 91 vs 58% > or =70% match percent, P < 0.001).

CONCLUSION

Wavelet-based morphology scores in ICDs may change with various body positions. These variations are relatively minor using the nominal configuration (RV coil-can), as well as by using RV coil-SVC coil and tip-ring. However, morphology scores can vary considerably when SVC coil-can is used; therefore, this configuration should be avoided while using the wavelet algorithm.