AI Filter Improves Positive Predictive Value of Atrial Fibrillation Detection by an Implantable Loop Recorder

Spontaneous fluctuation in atrial fibrillation burden and duration in patients with implantable loop monitors

BACKGROUND

Most studies of device-detected atrial fibrillation (AF) have recommended indefinite anticoagulation once a patient crosses a particular threshold for AF duration or burden. However, durations and burdens are known to fluctuate over time, but little is known about the magnitude of spontaneous fluctuations and the potential impact on anticoagulation decisions.

OBJECTIVE

To quantify spontaneous fluctuations in AF duration and burden in patients with implantable loop recorders (ILRs) METHODS: We reviewed all ILR interrogations for patients with non-permanent AF at our institution from 2018 to 2023. We excluded patients treated with rhythm control. The duration of longest AF episode at each interrogation was classified as < 6, 6-24, and > 24 h, and the AF burden reported at each interrogation was classified as < 2%, 2%-11.4%, and > 11.4%.

RESULTS

Out of 156 patients, the mean age at ILR implant was 70.9 ± 12.5 years, CHA2DS2-VASc score was 4.2 ± 1.8, duration of ILR follow-up was 23.4 ± 11.2 months, and number of ILR interrogations per patient was 18.0 ± 8.9. The duration of longest AF episode at any point during follow-up was < 6 , 6-24 , and > 24 h in 110, 30, and 16 patients, respectively. Among the 30 patients with a longest AF episode of 6-24 h at some point during follow-up, out of 594 total ILR interrogations, only 75 (12%) showed a longest episode of 6-24 h. In the remaining 519 interrogations, the longest episode was < 6 h. In patients with a longest episode of > 24 h at any point during follow-up (n = 16), only 47 out of 320 total ILR interrogations (15%) showed an episode of > 24 h. When evaluating AF burden, 96, 38, and 22 patients had maximum reported AF burdens of < 2%, 2%-11.4%, and > 11.4% at any point during ILR follow-up. Among those with a maximum burden of 2%-11.4% at some point during follow-up (n = 38), out of 707 ILR interrogations, only 76 (11%) showed a burden of 2%-11.4%. In the remaining 631 interrogations, the burden was < 2%. In the 22 patients with a burden > 11.4% at some point during follow-up, only 80 out of 480 interrogations (17%) showed a burden of > 11.4%. In 65% of interrogations, the burden was < 2%.

CONCLUSION

Significant, spontaneous fluctuations in AF burden and duration are common in patients with ILRs. Even in patients with AF episodes of 6-24 h or > 24 h at some point during follow-up, the vast majority of interrogations show episodes of < 6 h. Similarly, in patients with burdens of 2%-11.4% or > 11.4% at some point during follow-up, the vast majority of interrogations show burdens of < 2%. More data are needed to determine whether crossing an AF burden or duration threshold once is sufficient to merit lifelong anticoagulation or whether spontaneous fluctuations in AF burden and duration should impact anticoagulation decisions.

Use of Artificial Intelligence Including Multimodal Systems to Improve the Management of Cardiovascular Disease

The rising prevalence of cardiovascular disease presents an escalating challenge for current health services, which are grappling with increasing demands. Innovative changes are imperative to sustain the delivery of high-quality patient care. Recent technologic advances have resulted in the emergence of artificial intelligence as a viable solution. Advanced algorithms are now capable of performing complex analysis of large volumes of data rapidly and with exceptional accuracy. Multimodality artificial intelligence systems handle a diverse range of data including images, text, video, and audio. Compared with single-modality systems, multimodal artificial intelligence systems appear to hold promise for enhancing overall performance and enabling smoother integration into existing workflows. Such systems can empower physicians with clinical decision support and enhanced efficiency. Owing to the complexity of the field, however, truly multimodal artificial intelligence is still scarce in the management of cardiovascular disease. This article aims to cover current research, emerging trends, and the future utilisation of artificial intelligence in the management of cardiovascular disease, with a focus on multimodality systems.

Scientific statement from the French neurovascular and cardiac societies for improved detection of atrial fibrillation after ischaemic stroke and transient ischaemic attack

Atrial fibrillation (AF) is the primary cause of ischaemic stroke and transient ischaemic attack (TIA). AF is associated with a high risk of recurrence, which can be reduced using optimal prevention strategies, mainly anticoagulant therapy. The availability of effective prophylaxis justifies the need for a significant, coordinated and thorough transdisciplinary effort to screen for AF associated with stroke. A recent French national survey, initiated and supported by the Société française neurovasculaire (SFNV) and the Société française de cardiologie (SFC), revealed many shortcomings, such as the absence or inadequacy of telemetry equipment in more than half of stroke units, insufficient and highly variable access to monitoring tools, delays in performing screening tests, heterogeneous access to advanced or connected ambulatory monitoring techniques, and a lack of dedicated human resources. The present scientific document has been prepared on the initiative of the SFNV and the SFC with the aim of helping to address the current shortcomings and gaps, to promote efficient and cost-effective AF detection, and to improve and, where possible, homogenize the quality of practice in AF screening among stroke units and outpatient post-stroke care networks. The working group, composed of cardiologists and vascular neurologists who are experts in the field and are nominated by their peers, reviewed the literature to propose statements, which were discussed in successive cycles, and maintained, either by consensus or by vote, as appropriate. The text was then submitted to the SFNV and SFC board members for review. This scientific statement document argues for the widespread development of patient pathways to enable the most efficient AF screening after stroke. This assessment should be carried out by a multidisciplinary team, including expert cardiologists and vascular neurologists.

Detection of atrial fibrillation using a nonlinear Lorenz Scattergram and deep learning in primary care

BACKGROUND

Atrial fibrillation (AF) is highly correlated with heart failure, stroke and death. Screening increases AF detection and facilitates the early adoption of comprehensive intervention. Long-term wearable devices have become increasingly popular for AF screening in primary care. However, interpreting data obtained by long-term wearable ECG devices is a problem in primary care. To diagnose the disease quickly and accurately, we aimed to build AF episode detection model based on a nonlinear Lorenz scattergram (LS) and deep learning.

METHODS

The MIT-BIH Normal Sinus Rhythm Database, MIT-BIH Arrhythmia Database and the Long-Term AF Database were extracted to construct the MIT-BIH Ambulatory Electrocardiograph (MIT-BIH AE) dataset. We converted the long-term ECG into a two-dimensional LSs. The LSs from MIT-BIH AE dataset was randomly divided into training and internal validation sets in a 9:1 ratio, which was used to develop and internally validated model. We built a MOBILE-SCREEN-AF (MS-AF) dataset from a single-lead wearable ECG device in primary care for external validation. Performance was quantified using a confusion matrix and standard classification metrics.

RESULTS

During the evaluation of model performance based on the LS, the sensitivity, specificity and accuracy of the model in diagnosing AF were 0.992, 0.973, and 0.983 in the internal validation set respectively. In the external validation set, these metrics were 0.989, 0.956, and 0.967, respectively. Furthermore, when evaluating the model's performance based on ECG records in the MS-AF dataset, the sensitivity, specificity and accuracy of model diagnosis paroxysmal AF were 1.000, 0.870 and 0.876 respectively, and 0.927, 1.000 and 0.973 for the persistent AF.

CONCLUSIONS

The model based on the nonlinear LS and deep learning has high accuracy, making it promising for AF screening in primary care. It has potential for generalization and practical application.

Validation and clinical application of implantable loop recorders for diagnosis of atrial fibrillation in horses

BACKGROUND

Paroxysmal atrial fibrillation (pAF) occurs sporadically and can impair athletic performance. Gold standard for diagnosis is surface electrocardiography (ECG), however, this requires AF to be sustained. Implantable loop recorders (ILRs) are routinely used for AF detection in human medicine. While ILR placement has been studied in horses, its AF detection performance is unknown.

OBJECTIVES

(I) Validation of ILRs for AF detection in horses. (II) Determining pAF incidence using ILRs and estimate the positive predictive value (PPV).

STUDY DESIGN

(I) Experimental study; (II) Longitudinal observational study.

METHODS

(I) Implantation of ILRs in 15 horses with AF and 13 horses in sinus rhythm. Holter ECGs were recorded at: 1, 4, 8, 12 and 16 weeks of AF. The ILR ECGs were compared with surface ECGs to assess diagnostic sensitivity and specificity. (II) Eighty horses (43 Warmbloods, 37 Standardbreds) with ILRs were monitored for 367 days [IQR 208-621].

RESULTS

(I) ILRs detected AF on all recording days, in horses with AF, with a sensitivity of 66.1% (95% CI: 65.8-66.5) and a specificity of 99.99% (95% CI: 99.97-99.99). The sensitivity remained consistent across all time points. (II) The incidence of pAF was 6.3% (5/80). In horses with pAF, the PPV ranged from 8% to 87%. Increased body condition score (BCS > 6/9) was associated with an increased number of false positive episodes (p = 0.005).

MAIN LIMITATIONS

(I) Horses were stabled during the ECG recordings, and AF was induced, rather than naturally occurring pAF. (II) Integrated algorithm in this ILR is optimised for AF detection in humans using remote monitors. Additionally, sensing is affected by motion artefacts.

CONCLUSION

The ILR reliably detected AF in resting horses, particularly in horses with normal BCS (6/9). The ILR proved useful to detect pAF and is recommended alongside Holter monitoring for diagnostic workup of horses with suspected pAF.

Contemporary Advances in Cardiac Remote Monitoring: A Comprehensive, Updated Mini-Review

Over the past decade, remote monitoring (RM) has become an increasingly popular way to improve healthcare and health outcomes. Modern cardiac implantable electronic devices (CIEDs) are capable of recording an increasing amount of data related to CIED function, arrhythmias, physiological status and hemodynamic parameters, providing in-depth and updated information on patient cardiovascular function. The extensive use of RM for patients with CIED allows for early diagnosis and rapid assessment of relevant issues, both clinical and technical, as well as replacing outpatient follow-up improving overall management without compromise safety. This approach is recommended by current guidelines for all eligible patients affected by different chronic cardiac conditions including either brady- and tachy-arrhythmias and heart failure. Beyond to clinical advantages, RM has demonstrated cost-effectiveness and is associated with elevated levels of patient satisfaction. Future perspectives include improving security, interoperability and diagnostic power as well as to engage patients with digital health technology. This review aims to update existing data concerning clinical outcomes in patients managed with RM in the wide spectrum of cardiac arrhythmias and Hear Failure (HF), disclosing also about safety, effectiveness, patient satisfaction and cost-saving.

Monitoring of Remotely Reprogrammable Implantable Loop Recorders With Algorithms to Reduce False-Positive Alerts

BACKGROUND

Implantable loop recorders (ILRs) are increasingly placed for arrhythmia detection. However, historically, ≈75% of ILR alerts are false positives, requiring significant time and effort for adjudication. The LINQII and LUX-Dx are remotely reprogrammable ILRs with dual-stage algorithms using artificial intelligence to reduce false positives, but their utility in routine clinical practice has not been studied.

METHODS AND RESULTS

We identified patients with the LINQII and LUX-Dx who were monitored by the Veterans Affairs National Cardiac Device Surveillance Program between March and June 2022. ILR programming was customized on the basis of implant indication. All alerts and every 90-day scheduled transmissions were manually reviewed. ILRs were remotely reprogrammed, as appropriate, after false-positive alerts or 2 consecutive same-type alerts, unless there was ongoing clinical need for that alert. Outcomes were total number of transmissions and false positives. We performed medical record review to determine if patients experienced any adverse clinical events, including hospitalization and mortality. Among 117 LINQII patients, there were 239 total alerts, 43 (18.0%) of which were false positives. Among 105 LUX-Dx patients, there were 300 total alerts, 115 (38.3%) of which were false positives. LINQIIs were reprogrammed 22 times, resulting in a decrease in median alerts/day from 0.13 to 0.03. LUX-Dx ILRs were reprogrammed 52 times, resulting in a decrease from 0.15 to 0.01 median alerts/day. There were no adverse clinical events that could have been identified by superior or earlier arrhythmia detection.

CONCLUSIONS

ILRs with artificial intelligence algorithms and remote reprogramming ability are associated with reduced alert burden because of higher true-positive rates than prior ILRs, without missing potentially consequential arrhythmias.

False-positive alarms in patients with implantable loop recorder followed by remote monitoring: A systematic review

Remote Monitoring (RM) has been shown to provide useful information about arrhythmic events in patients with implantable loop recorders (ILRs), however there is few and conflicting data about the false positive (FP) alarms burden and characteristics among ILR recipients. The aim of the present systematic review was to evaluate incidence and characteristics of FP alarms among ILR patients followed by RM. We developed a systematic research in Embase, MEDLINE and PubMed databases and selected all papers focused on false positive ILR transmissions published from June 1, 2013 to June 1, 2023. Case reports, meeting summaries, posters and simple reviews were excluded. Twelve reports were finally selected, including five prospective and seven retrospective studies. Information about population characteristics, device type and setting, overall transmissions and FP alarms and any adopted strategies to reduce them were extracted from an overall population of 3.305 patients. FP alarms were 59.7% of the overall remote transmissions and were found in 1/5 of the analyzed population. FP alarms for atrial fibrillation were the most common cause of false transmissions and were mainly due to premature atrial and ventricular complexes. No clinical predictors of FP alarms were identified, except for nonparasternal ILR implantation site. Since the overload work due to FP alarms might reduce the benefit of remote monitoring of ILR patients, the device optimization is an important step until an help from machine-learning algorithms is available.

Detection of paroxysmal atrial fibrillation preceding persistent atrial fibrillation in a horse using an implantable loop recorder with remote monitoring

Implantable loop recorders (ILRs) are increasingly used in equine cardiology to detect arrhythmias in the context of collapse, poor performance or monitoring for recurrence of atrial fibrillation (AF). However to date, the ILR has never been reported to be used with a remote monitoring functionality in horses, therefore the arrhythmia is only discovered when a clinician interrogates the ILR using dedicated equipment, which might delay diagnosis and intervention. This case report describes the use of an ILR with remote monitoring functionality in a horse with recurrent AF. The remote monitoring consisted of a transmission device located in the stable allowing daily transmission of arrhythmia recordings and functioning messages to an online server, available for the clinician to evaluate without specialised equipment. The ILR detected an episode of paroxysmal AF approximately 3 months after implantation. Seven months after implantation, initiation of persistent AF was seen on an episode misclassified by the ILR as bradycardia, and the horse was retired. This report shows the feasibility and benefits of remote monitoring for ILRs in horses, but also the shortcomings of current algorithms to interpret the equine electrocardiogram.

Implantable cardiac monitors: artificial intelligence and signal processing reduce remote ECG review workload and preserve arrhythmia detection sensitivity

Introduction

Implantable cardiac monitors (ICMs) provide long-term arrhythmia monitoring, but high rates of false detections increase the review burden. The new "SmartECG" algorithm filters false detections. Using large real-world data sets, we aimed to quantify the reduction in workload and any loss in sensitivity from this new algorithm.

Methods

Patients with a BioMonitor IIIm and any device indication were included from three clinical projects. All subcutaneous ECGs (sECGs) transmitted via remote monitoring were classified by the algorithm as "true" or "false." We quantified the relative reduction in workload assuming "false" sECGs were ignored. The remote monitoring workload from five hospitals with established remote monitoring routines was evaluated. Loss in sensitivity was estimated by testing a sample of 2000 sECGs against a clinical board of three physicians.

Results

Of our population of 368 patients, 42% had an indication for syncope or pre-syncope and 31% for cryptogenic stroke. Within 418.5 patient-years of follow-up, 143,096 remote monitoring transmissions contained 61,517 sECGs. SmartECG filtered 42.8% of all sECGs as "false," reducing the number per patient-year from 147 to 84. In five hospitals, nine trained reviewers inspected on average 105 sECGs per working hour. This results in an annual working time per patient of 83 min without SmartECG, and 48 min with SmartECG. The loss of sensitivity is estimated as 2.6%. In the majority of cases where true arrhythmias were rejected, SmartECG classified the same type of arrhythmia as "true" before or within 3 days of the falsely rejected sECG.

Conclusion

SmartECG increases efficiency in long-term arrhythmia monitoring using ICMs. The reduction of workload by SmartECG is meaningful and the risk of missing a relevant arrhythmia due to incorrect filtering by the algorithm is limited.

A mutually communicable external system resource in remote monitoring for cardiovascular implantable electronic devices

BACKGROUND

Using third-party resources to manage remote monitoring (RM) data from implantable cardiac electronic devices (CIEDs) can assist in device clinic workflows. However, each hospital-acquired data is not used for further analysis as big data.

METHODS AND RESULTS

We developed a real-time and automatically centralized system of CIED information from multiple hospitals. If the extensive data-based analysis suggests individual problems, it can be returned to each hospital. To show its feasibility, we prospectively analyzed data from six hospitals. For example, unexpected abnormal battery levels were easily illustrated without recall information.

CONCLUSIONS

The centralized RM system could be a new platform that promotes the utilization of device data as big data, and that information could be used for each patient's practice.

Artificial intelligence cloud platform improves arrhythmia detection from insertable cardiac monitors to 25 cardiac rhythm patterns through multi-label classification

BACKGROUND

Electrocardiogram (ECG) is the gold standard for the diagnosis of cardiac arrhythmias and other heart diseases. Insertable cardiac monitors (ICMs) have been developed to continuously monitor cardiac activity over long periods of time and to detect 4 cardiac patterns (atrial tachyarrhythmias, ventricular tachycardia, bradycardia, and pause). However, interpretation of ECG or ICM subcutaneous ECG (sECG) is time-consuming for clinicians. Artificial intelligence (AI) classifies ECG and sECG with high accuracy in short times.

OBJECTIVE

To demonstrate whether an AI algorithm can expand ICM arrhythmia recognition from 4 to many cardiac patterns.

METHODS

We performed an exploratory retrospective study with sECG raw data coming from 20 patients wearing a Confirm Rx™ (Abbott, Sylmar, USA) ICM. The sECG data were recorded in standard conditions and then analyzed by AI (Willem™, IDOVEN, Madrid, Spain) and cardiologists, in parallel.

RESULTS

In nineteen patients, ICMs recorded 2261 sECGs in an average follow-up of 23 months. Within these 2261 sECG episodes, AI identified 7882 events and classified them according to 25 different cardiac rhythm patterns with a pondered global accuracy of 88%. Global positive predictive value, sensitivity, and F1-score were 86.77%, 83.89%, and 85.52% respectively. AI was especially sensitive for bradycardias, pauses, rS complexes, premature atrial contractions, and inverted T waves, reducing the median time spent to classify each sECG compared to cardiologists.

CONCLUSION

AI can process sECG raw data coming from ICMs without previous training, extending the performance of these devices and saving cardiologists' time in reviewing cardiac rhythm patterns detection.

Daily and automatic remote monitoring of implantable cardiac monitors: A descriptive analysis of transmitted episodes

BACKGROUND

Remote Monitoring (RM) is recognized for its ability to enhance the clinical management of patients with implantable cardiac monitor (ICM). This study aims to provide a comprehensive description of the arrhythmic episodes transmitted by a daily and automatic RM system from a cohort of ICM patients.

METHODS

The study retrospectively analyzed daily transmissions from consecutive patients who had been implanted with a long-sensing vector ICM (BIOMONITOR III/IIIm) at four sites. All transmitted arrhythmic recordings were evaluated to determine whether they were true positive episodes or false positives (FP).

RESULTS

A total of 14,136 episodes were transmitted from 119 patients (74.8% male, median age 62 years old) during a median follow-up of 371 days. The rate of arrhythmic episodes was 14.2 per patient-year (interquartile range: 1.8-126), with 97 patients (81.5%) experiencing at least one ICM activation. Fifty-five percent of episodes were identified as FP, and 67 patients (56.3%) had at least one inappropriate activation. The FP rate was 1.4 per patient-year (0-40). The best per-episode predictive positive values were observed for bradycardia and atrial fibrillation (0.595 and 0.553, respectively). Notably, the implementation of an algorithm designed to minimize false detections significantly reduced the prevalence of atrial fibrillation FP episodes (17.6% vs. 43.5%, p = 0.008).

CONCLUSION

Daily and automatic RM appears to be a reliable tool for the comprehensive remote management of ICM patients. However, the number of arrhythmic episodes requiring review is high, and further improvements are needed to reduce FP and facilitate accurate interpretation of transmissions.

Artificial intelligence for detection of ventricular oversensing: Machine learning approaches for noise detection within nonsustained ventricular tachycardia episodes remotely transmitted by pacemakers and implantable cardioverter-defibrillators

BACKGROUND

Pacemakers (PMs) and implantable cardioverter-defibrillators (ICDs) increasingly automatically record and remotely transmit nonsustained ventricular tachycardia (NSVT) episodes, which may reveal ventricular oversensing.

OBJECTIVES

We aimed to develop and validate a machine learning algorithm that accurately classifies NSVT episodes transmitted by PMs and ICDs in order to lighten health care workload burden and improve patient safety.

METHODS

PMs or ICDs (Boston Scientific, St Paul, MN) from 4 French hospitals with ≥1 transmitted NSVT episode were split into 3 subgroups: training set, validation set, and test set. Each NSVT episode was labeled as either physiological or nonphysiological. Four machine learning algorithms-2DTF-CNN, 2D-DenseNet, 2DTF-VGG, and 1D-AgResNet-were developed using training and validation data sets. Accuracies of the classifiers were compared with an analysis of the remote monitoring team of the Bordeaux University Hospital using F2 scores (favoring sensitivity over predictive positive value) using an independent test set.

RESULTS

A total of 807 devices transmitted 10,471 NSVT recordings (82% ICD; 18% PM), of which 87 devices (10.8%) transmitted 544 NSVT recordings with nonphysiological signals. The classification by the remote monitoring team resulted in an F2 score of 0.932 (sensitivity 95%; specificity 99%) The 4 machine learning algorithms showed high and comparable F2 scores (2DTF-CNN: 0.914; 2D-DenseNet: 0.906; 2DTF-VGG: 0.863; 1D-AgResNet: 0.791), and only 1D-AgResNet had significantly different labeling from that of the remote monitoring team.

CONCLUSION

Machine learning algorithms were accurate in detecting nonphysiological signals within electrograms transmitted by PMs and ICDs. An artificial intelligence approach may render remote monitoring less resourceful and improve patient safety.

2023 HRS/EHRA/APHRS/LAHRS expert consensus statement on practical management of the remote device clinic

Remote monitoring is beneficial for the management of patients with cardiovascular implantable electronic devices by impacting morbidity and mortality. With increasing numbers of patients using remote monitoring, keeping up with higher volume of remote monitoring transmissions creates challenges for device clinic staff. This international multidisciplinary document is intended to guide cardiac electrophysiologists, allied professionals, and hospital administrators in managing remote monitoring clinics. This includes guidance for remote monitoring clinic staffing, appropriate clinic workflows, patient education, and alert management. This expert consensus statement also addresses other topics such as communication of transmission results, use of third-party resources, manufacturer responsibilities, and programming concerns. The goal is to provide evidence-based recommendations impacting all aspects of remote monitoring services. Gaps in current knowledge and guidance for future research directions are also identified.

Novel algorithms improve arrhythmia detection accuracy in insertable cardiac monitors

INTRODUCTION

Insertable cardiac monitors (ICMs) are commonly used to diagnose cardiac arrhythmias. False detections in the latest ICM systems remain an issue, primarily due to inaccurate R-wave sensing. New discrimination algorithms were developed and tested to reduce false detections of atrial fibrillation (AF), pause, and tachycardia episodes in ICMs.

METHODS

Stored electrograms (EGMs) of AF, pause, and tachycardia episodes detected by Abbott Confirm Rx™ ICMs were extracted from the Merlin.net™ Patient Care Network, and manually adjudicated to establish independent training and testing datasets. New discrimination algorithms were developed to reject false episodes due to inaccurate R-wave sensing, P-wave identification, and R-R interval patterns. The performance of these new algorithms was quantified by false positive reduction (FPR) and true positive maintenance (TPM), relative to the existing algorithms.

RESULTS

The new AF detection algorithm was trained on 5911 EGMs from 744 devices, resulting in 66.9% FPR and 97.8% TPM. In the testing data set of 1354 EGMs from 119 devices, this algorithm achieved 45.8% FPR and 97.0% TPM. The new pause algorithm was trained on 7178 EGMs from 1490 devices, resulting in 70.9% FPR and 98.7% TPM. In the testing data set of 1442 EGMs from 340 devices, this algorithm achieved 74.4% FPR and 99.3% TPM. The new tachycardia algorithm was trained on 520 EGMs from 204 devices, resulting in 57.0% FPR and 96.6% TPM. In the testing data set of 459 EGMs from 237 devices, this algorithm achieved 57.9% FPR and 96.5% TPM.

CONCLUSION

The new algorithms substantially reduced false AF, pause, and tachycardia episodes while maintaining the majority of true arrhythmia episodes detected by the Abbott ICM algorithms that exist today. Implementing these algorithms in the next-generation ICM systems may lead to improved detection accuracy, in-clinic efficiency, and device battery longevity.

The digital journey: 25 years of digital development in electrophysiology from an Europace perspective

AIMS

Over the past 25 years there has been a substantial development in the field of digital electrophysiology (EP) and in parallel a substantial increase in publications on digital cardiology.In this celebratory paper, we provide an overview of the digital field by highlighting publications from the field focusing on the EP Europace journal.

RESULTS

In this journey across the past quarter of a century we follow the development of digital tools commonly used in the clinic spanning from the initiation of digital clinics through the early days of telemonitoring, to wearables, mobile applications, and the use of fully virtual clinics. We then provide a chronicle of the field of artificial intelligence, a regulatory perspective, and at the end of our journey provide a future outlook for digital EP.

CONCLUSION

Over the past 25 years Europace has published a substantial number of papers on digital EP, with a marked expansion in digital publications in recent years.

Remote monitoring of implantable loop recorders reduces time to diagnosis in patients with unexplained syncope: a multicenter propensity score-matched study

Background

There are little data on remote monitoring (RM) of implantable loop recorders (ILRs) in patients with unexplained syncope and whether it confers enhanced diagnostic power.

Objective

To evaluate the effect of RM in ILR recipients for unexplained syncope for early detection of clinically relevant arrhythmias by comparison with a historical cohort with no RM.

Methods

SyncRM is a propensity score (PS)-matched study prospectively including 133 consecutive patients with unexplained syncope and ILR followed up by RM (RM-ON group). A historical cohort of 108 consecutive ILR patients with biannual in-hospital follow-up visits was used as control group (RM-OFF group). The primary endpoint was the time to the clinician's evaluation of clinically relevant arrhythmias (types 1, 2, and 4 of the ISSUE classification).

Results

The primary endpoint of arrhythmia evaluation was reached in 38 patients (28.6%) of the RM-ON group after a median time of 46 days (interquartile range, 13-106) and in 22 patients (20.4%) of the RM-OFF group after 92 days (25-368). The PS-matched adjusted ratio of rates of arrhythmia evaluation was 2.53 (95% confidence interval, 1.32-4.86) in the RM-ON vs. RM-OFF group (p = 0.005).

Conclusion

In our PS-matched comparison with a historical cohort, RM of ILR patients with unexplained syncope was associated with a 2.5-fold higher chance of evaluations of clinically relevant arrhythmias as compared with biannual in-office follow-up visits.

2023 HRS/EHRA/APHRS/LAHRS Expert Consensus Statement on Practical Management of the Remote Device Clinic

Remote monitoring is beneficial for the management of patients with cardiovascular implantable electronic devices by impacting morbidity and mortality. With increasing numbers of patients using remote monitoring, keeping up with higher volume of remote monitoring transmissions creates challenges for device clinic staff. This international multidisciplinary document is intended to guide cardiac electrophysiologists, allied professionals, and hospital administrators in managing remote monitoring clinics. This includes guidance for remote monitoring clinic staffing, appropriate clinic workflows, patient education, and alert management. This expert consensus statement also addresses other topics such as communication of transmission results, use of third-party resources, manufacturer responsibilities, and programming concerns. The goal is to provide evidence-based recommendations impacting all aspects of remote monitoring services. Gaps in current knowledge and guidance for future research directions are also identified.

2023 HRS/EHRA/APHRS/LAHRS Expert Consensus Statement on Practical Management of the Remote Device Clinic

Remote monitoring is beneficial for the management of patients with cardiovascular implantable electronic devices by impacting morbidity and mortality. With increasing numbers of patients using remote monitoring, keeping up with higher volume of remote monitoring transmissions creates challenges for device clinic staff. This international multidisciplinary document is intended to guide cardiac electrophysiologists, allied professionals, and hospital administrators in managing remote monitoring clinics. This includes guidance for remote monitoring clinic staffing, appropriate clinic workflows, patient education, and alert management. This expert consensus statement also addresses other topics such as communication of transmission results, use of third-party resources, manufacturer responsibilities, and programming concerns. The goal is to provide evidence-based recommendations impacting all aspects of remote monitoring services. Gaps in current knowledge and guidance for future research directions are also identified.

Clinical validation of an artificial intelligence algorithm offering cross-platform detection of atrial fibrillation using smart device electrocardiograms

BACKGROUND

Several smart devices are able to detect atrial fibrillation automatically by recording a single-lead electrocardiogram, and have created a work overload at the hospital level as a result of the need for over-reads by physicians.

AIM

To compare the atrial fibrillation detection performances of the manufacturers' algorithms of five smart devices and a novel deep neural network-based algorithm.

METHODS

We compared the rate of inconclusive tracings and the diagnostic accuracy for the detection of atrial fibrillation between the manufacturers' algorithms and the deep neural network-based algorithm on five smart devices, using a physician-interpreted 12-lead electrocardiogram as the reference standard.

RESULTS

Of the 117 patients (27% female, median age 65 years, atrial fibrillation present at time of recording in 30%) included in the final analysis (resulting in 585 analyzed single-lead electrocardiogram tracings), the deep neural network-based algorithm exhibited a higher conclusive rate relative to the manufacturer algorithm for all five models: 98% vs. 84% for Apple; 99% vs. 81% for Fitbit; 96% vs. 77% for AliveCor; 99% vs. 85% for Samsung; and 97% vs. 74% for Withings (P<0.01, for each model). When applying our deep neural network-based algorithm, sensitivity and specificity to correctly identify atrial fibrillation were not significantly different for all assessed smart devices.

CONCLUSION

In this clinical validation, the deep neural network-based algorithm significantly reduced the number of tracings labeled inconclusive, while demonstrating similarly high diagnostic accuracy for the detection of atrial fibrillation, thereby providing a possible solution to the data surge created by these smart devices.

A two-staged classifier to reduce false positives: On device detection of atrial fibrillation using phase-based distribution of poincaré plots and deep learning

BACKGROUND

Mobile Cardiac Outpatient Telemetry (MCOT) can be used to screen high risk patients for atrial fibrillation (AF). These devices rely primarily on algorithmic detection of AF events, which are then stored and transmitted to a clinician for review. It is critical the positive predictive value (PPV) of MCOT detected AF is high, and this often leads to reduced sensitivity, as device manufacturers try to limit false positives.

OBJECTIVE

The purpose of this study was to design a two stage classifier using artificial intelligence (AI) to improve the PPV of MCOT detected atrial fibrillation episodes whilst maintaining high levels of detection sensitivity.

METHODS

A low complexity, RR-interval based, AF classifier was paired with a deep convolutional neural network (DCNN) to create a two-stage classifier. The DCNN was limited in size to allow it to be embedded on MCOT devices. The DCNN was trained on 491,727 ECGs from a proprietary database and contained 128,612 parameters requiring only 158 KB of storage. The performance of the two-stage classifier was then assessed using publicly available datasets.

RESULTS

The sensitivity of AF detected by the low complexity classifier was high across all datasets (>93%) however the PPV was poor (<76%). Subsequent analysis by the DCNN increased episode PPV across all datasets substantially (>11%), with only a minor loss in sensitivity (<5%). This increase in PPV was due to a decrease in the number of false positive detections. Further analysis showed that DCNN processing was only required on around half of analysis windows, offering a significant computational saving against using the DCNN as a one-stage classifier.

CONCLUSION

DCNNs can be combined with existing MCOT classifiers to increase the PPV of detected AF episodes. This reduces the review burden for physicians and can be achieved with only a modest decrease in sensitivity.

Implantable loop recorders in patients with atrial fibrillation

INTRODUCTION

Implantable loop recorders (ILRs) provide practitioners with high-quality electrocardiographic data over an extended monitoring period. These data can guide the diagnosis and management of patients with atrial fibrillation (AF).

AREAS COVERED

This review summarizes the available evidence and consensus statements supporting the use of ILRs in the detection of AF, as well as monitoring of patients with known AF. Future directions for research are also discussed.

EXPERT OPINION

ILRs are the gold standard for detecting AF, providing superior diagnostic yield compared to other modes of ambulatory electrocardiography monitoring. Both experimental evidence and consensus statements support the use of ILRs in clinical settings where the diagnosis of AF may significantly change management, or where a high degree of sensitivity is needed. ILRs may also be used to monitor patients following AF ablation. More evidence is needed to better inform how ILR-detected AF should change management.

An ensemble of features based deep learning neural network for reduction of inappropriate atrial fibrillation detection in implantable cardiac monitors

Background

Multiple studies have reported on classification of raw electrocardiograms (ECGs) using convolutional neural networks (CNNs).

Objective

We investigated an application-specific CNN using a custom ensemble of features designed based on characteristics of the ECG during atrial fibrillation (AF) to reduce inappropriate AF detections in implantable cardiac monitors (ICMs).

Methods

An ensemble of features was developed and combined to form an input signal for the CNN. The features were based on the morphological characteristics of AF, incoherence of RR intervals, and the fact that AF begets more AF. A custom CNN model and the RESNET18 model were trained using ICM-detected AF episodes that were adjudicated to be true AF or false detections. The trained models were evaluated using a test dataset from independent patients.

Results

The training and validation datasets consisted of 31,757 AF episodes (2516 patients) and 28,506 false episodes (2126 patients). The validation set (20% randomly chosen episodes of each type) had an area under the curve of 0.996 for custom CNN (0.993 for RESNET18). Thresholds were chosen to obtain a relative sensitivity and specificity of 99.2% and 92.8%, respectively (99.2% and 87.9% for RESNET18, respectively). The performance in the independent test set (4546 AF episodes from 418 patients; 5384 false episodes from 605 patients) showed an area under the curve of 0.993 (0.991 for RESNET18) and relative sensitivity and specificity of 98.7% and 91.4%, respectively, at chosen thresholds (98.9% and 88.2% for RESNET18, respectively).

Conclusion

An ensemble of features-based CNNs was developed that reduced inappropriate AF detection in ICMs by over 90% while preserving sensitivity.

Evaluation of an Ambulatory ECG Analysis Platform Using Deep Neural Networks in Routine Clinical Practice

Background Holter analysis requires significant clinical resources to achieve a high-quality diagnosis. This study sought to assess whether an artificial intelligence (AI)-based Holter analysis platform using deep neural networks is noninferior to a conventional one used in clinical routine in detecting a major rhythm abnormality. Methods and Results A total of 1000 Holter (24-hour) recordings were collected from 3 tertiary hospitals. Recordings were independently analyzed by cardiologists for the AI-based platform and by electrophysiologists as part of clinical practice for the conventional platform. For each Holter, diagnostic performance was evaluated and compared through the analysis of the presence or absence of 5 predefined cardiac abnormalities: pauses, ventricular tachycardia, atrial fibrillation/flutter/tachycardia, high-grade atrioventricular block, and high burden of premature ventricular complex (>10%). Analysis duration was monitored. The deep neural network-based platform was noninferior to the conventional one in its ability to detect a major rhythm abnormality. There were no statistically significant differences between AI-based and classical platforms regarding the sensitivity and specificity to detect the predefined abnormalities except for atrial fibrillation and ventricular tachycardia (atrial fibrillation, 0.98 versus 0.91 and 0.98 versus 1.00; pause, 0.95 versus 1.00 and 1.00 versus 1. 00; premature ventricular contractions, 0.96 versus 0.87 and 1.00 versus 1.00; ventricular tachycardia, 0.97 versus 0.68 and 0.99 versus 1.00; atrioventricular block, 0.93 versus 0.57 and 0.99 versus 1.00). The AI-based analysis was >25% faster than the conventional one (4.4 versus 6.0 minutes; P<0.001). Conclusions These preliminary findings suggest that an AI-based strategy for the analysis of Holter recordings is faster and at least as accurate as a conventional analysis by electrophysiologists.

ESC Working Group on e-Cardiology Position Paper: accuracy and reliability of electrocardiogram monitoring in the detection of atrial fibrillation in cryptogenic stroke patients : In collaboration with the Council on Stroke, the European Heart Rhythm Association, and the Digital Health Committee

The role of subclinical atrial fibrillation as a cause of cryptogenic stroke is unambiguously established. Long-term electrocardiogram (ECG) monitoring remains the sole method for determining its presence following a negative initial workup. This position paper of the European Society of Cardiology Working Group on e-Cardiology first presents the definition, epidemiology, and clinical impact of cryptogenic ischaemic stroke, as well as its aetiopathogenic association with occult atrial fibrillation. Then, classification methods for ischaemic stroke will be discussed, along with their value in providing meaningful guidance for further diagnostic efforts, given disappointing findings of studies based on the embolic stroke of unknown significance construct. Patient selection criteria for long-term ECG monitoring, crucial for determining pre-test probability of subclinical atrial fibrillation, will also be discussed. Subsequently, the two major classes of long-term ECG monitoring tools (non-invasive and invasive) will be presented, with a discussion of each method's pitfalls and related algorithms to improve diagnostic yield and accuracy. Although novel mobile health (mHealth) devices, including smartphones and smartwatches, have dramatically increased atrial fibrillation detection post ischaemic stroke, the latest evidence appears to favour implantable cardiac monitors as the modality of choice; however, the answer to whether they should constitute the initial diagnostic choice for all cryptogenic stroke patients remains elusive. Finally, institutional and organizational issues, such as reimbursement, responsibility for patient management, data ownership, and handling will be briefly touched upon, despite the fact that guidance remains scarce and widespread clinical application and experience are the most likely sources for definite answers.

Device agnostic AI-based analysis of ambulatory ECG recordings

Deep Convolutional Neural Networks (DCNNs) have been shown to provide improved performance over traditional heuristic algorithms for the detection of arrhythmias from ambulatory ECG recordings. However, these DCNNs have primarily been trained and tested on device-specific databases with standardized electrode positions and uniform sampling frequencies. This work explores the possibility of training a DCNN for Atrial Fibrillation (AF) detection on a database of single‑lead ECG rhythm strips extracted from resting 12‑lead ECGs. We then test the performance of the DCNN on recordings from ambulatory ECG devices with different recording leads and sampling frequencies. We developed an extensive proprietary resting 12‑lead ECG dataset of 549,211 patients. This dataset was randomly split into a training set of 494,289 patients and a testing set of the remaining 54,922 patients. We trained a 34-layer convolutional DCNN to detect AF and other arrhythmias on this dataset. The DCNN was then validated on two Physionet databases commonly used to benchmark automated ECG algorithms (1) MIT-BIH Arrhythmia Database and (2) MIT-BIH Atrial Fibrillation Database. Validation was performed following the EC57 guidelines, with performance assessed by gross episode and duration sensitivity and positive predictive value (PPV). Finally, validation was also performed on a selection of rhythm strips from an ambulatory ECG patch that a committee of board-certified cardiologists annotated. On MIT-BIH, The DCNN achieved a sensitivity of 100% and 84% PPV in detecting episodes of AF. and 100% sensitivity and 94% PPV in quantifying AF episode duration. On AFDB, The DCNN achieved a sensitivity of 94% and PPV of 98% in detecting episodes of AF, and 98% sensitivity and 100% PPV in quantifying AF episode duration. On the patch database, the DCNN demonstrated performance that was closely comparable to that of a cardiologist. The results indicate that DCNN models can learn features that generalize between resting 12‑lead and ambulatory ECG recordings, allowing DCNNs to be device agnostic for detecting arrhythmias from single‑lead ECG recordings and enabling a range of clinical applications.

Artificial intelligence augments detection accuracy of cardiac insertable cardiac monitors: Results from a pilot prospective observational study

Background

Insertable cardiac monitors (ICMs) are indicated for long-term monitoring of patients with unexplained syncope or who are at risk for cardiac arrhythmias. The volume of ICM-transmitted information may result in long data review times to identify true and clinically relevant arrhythmias.

Objective

The purpose of this study was to evaluate whether artificial intelligence (AI) may improve ICM detection accuracy.

Methods

We performed a retrospective analysis of consecutive patients implanted with the Confirm Rx^TM ICM (Abbott) and followed in a prospective observational study. This device continuously monitors subcutaneous electrocardiograms (SECGs) and transmits to clinicians information about detected arrhythmias and patient-activated symptomatic episodes. All SECGs were classified by expert electrophysiologists and by the Willem^TM AI algorithm (IDOVEN).

Results

During mean follow-up of 23 months, of 20 ICM patients (mean age 68 ± 12 years; 50% women), 19 had 2261 SECGs recordings associated with cardiac arrhythmia detections or patient symptoms. True arrhythmias occurred in 11 patients: asystoles in 2, bradycardias in 3, ventricular tachycardias in 4, and atrial tachyarrhythmias (atrial tachycardia/atrial fibrillation [AT/AF]) in 10; with 6 patients having >1 arrhythmia type. AI algorithm overall accuracy for arrhythmia classification was 95.4%, with 97.19% sensitivity, 94.52% specificity, 89.74% positive predictive value, and 98.55% negative predictive value. Application of AI would have reduced the number of false-positive results by 98.0% overall: 94.0% for AT/AF, 87.5% for ventricular tachycardia, 99.5% for bradycardia, and 98.8% for asystole.

Conclusion

Application of AI to ICM-detected episodes is associated with high classification accuracy and may significantly reduce health care staff workload by triaging ICM data.

Wearables, telemedicine, and artificial intelligence in arrhythmias and heart failure: Proceedings of the European Society of Cardiology: Cardiovascular Round Table

Digital technology is now an integral part of medicine. Tools for detecting, screening, diagnosis, and monitoring health-related parameters have improved patient care and enabled individuals to identify issues leading to better management of their own health. Wearable technologies have integrated sensors and can measure physical activity, heart rate and rhythm, and glucose and electrolytes. For individuals at risk, wearables or other devices may be useful for early detection of atrial fibrillation or sub-clinical states of cardiovascular disease, disease management of cardiovascular diseases such as hypertension and heart failure, and lifestyle modification. Health data are available from a multitude of sources, namely clinical, laboratory and imaging data, genetic profiles, wearables, implantable devices, patient-generated measurements, and social and environmental data. Artificial intelligence is needed to efficiently extract value from this constantly increasing volume and variety of data and to help in its interpretation. Indeed, it is not the acquisition of digital information, but rather the smart handling and analysis that is challenging. There are multiple stakeholder groups involved in the development and effective implementation of digital tools. While the needs of these groups may vary, they also have many commonalities, including the following: a desire for data privacy and security; the need for understandable, trustworthy, and transparent systems; standardized processes for regulatory and reimbursement assessments; and better ways of rapidly assessing value.

Accuracy of atrial fibrillation detection by an insertable cardiac monitor in patients undergoing catheter ablation: Results of the BioVAD study

BACKGROUND

Insertable cardiac monitors (ICMs) are increasingly used to evaluate the atrial fibrillation (AF) burden after catheter ablation of AF. BioMonitor III (BM3) is an ICM with a long sensing vector, which enhances sensing capabilities. The AF detection algorithm of the BM3 is based on R-R interval variability.

OBJECTIVE

To evaluate the performance of the AF detection algorithm of BM3 in patients before and after catheter ablation of AF using simultaneous Holter recordings.

METHODS

In this prospective study, we enrolled patients scheduled for catheter ablation of paroxysmal or persistent AF. After BM3 implantation, patients had a 4 days Holter registration before and 3 months after ablation. All true AF episodes ≥2 min on the Holter were annotated and matched with BM3 detected AF detections.

RESULTS

Thirty-one patients were enrolled (mean age 60 ± 8, 74% male, 68% paroxysmal AF). Fifty-six Holter registrations were performed in 30 patients. Twelve patients demonstrated at least one true AF episode with a total AF duration of 570 h. The AF burden accuracy of BM3 before catheter ablation was 99.6%, with a duration sensitivity of 98.6% and a duration specificity of 99.9%. The AF burden accuracy of BM3 after catheter ablation was 99.8%, with a duration sensitivity of 90.2% and a duration specificity of 99.9%. Overall, the AF burden detected on the Holter and BM3 demonstrated a high Pearson correlation coefficient of 0.996.

CONCLUSION

BM3 accurately detects AF burden in patients before and after catheter ablation of AF.

Artificial intelligence for the detection, prediction, and management of atrial fibrillation

The present article reviews the state of the art of machine learning algorithms for the detection, prediction, and management of atrial fibrillation (AF), as well as of the development and evaluation of artificial intelligence (AI) in cardiology and beyond. Today, AI detects AF with a high accuracy using 12-lead or single-lead electrocardiograms or photoplethysmography. The prediction of paroxysmal or future AF currently operates at a level of precision that is too low for clinical use. Further studies are needed to determine whether patient selection for interventions may be possible with machine learning.

Artificial intelligence and cardiology: Current status and perspective: Artificial Intelligence and Cardiology

The development of artificial intelligence (AI) began in the mid-20th century but has been rapidly accelerating in the past decade. Reflecting the development of digital health over the past few years, this trend is also seen in medicine. The field of cardiovascular medicine uses a wide variety and a large amount of biosignals, so there are many situations where AI can contribute. The development of AI is in progress for all aspects of the healthcare system, including the prevention, screening, and treatment of diseases and the prediction of the prognosis. AI is expected to be used to provide specialist-level medical care, even in a situation where medical resources are scarce. However, like other medical devices, the concept and mechanism of AI must be fully understood when used; otherwise, it may be used inappropriately, resulting in detriment to the patient. Therefore, it is important to understand what we need to know as a cardiologist handling AI. This review introduces the basics and principles of AI, then shows how far the current development of AI has come, and finally gives a brief introduction of how to start the AI development for those who want to develop their own AI.