State of the art of mobile health technologies use in clinical arrhythmia care

Arrhythmias including AF, PAC, PVC, SVT, and VT are usually paroxysmal (suddenly increase or recur). The critical step of establishing symptom-rhythm correlation has required the use of medical-grade ambulatory electrocardiogram (ECG) monitors, 24–48 Holter, or longer-term event recorders. These tools have been limited in the time they can monitor a patient, which is problematic if no symptoms are experienced while wearing the monitor. mHealth devices, particularly smartphone-based ECG and PPG technology, facilitate relatively inexpensive, long-term rhythm monitoring. This can enable the patient themselves to detect arrhythmias, albeit with some limitations. In this section, we discuss the evidence that these devices can be used to screen for arrhythmias, confirm diagnosis, and assess response to treatment, including medical or interventional procedures, such as catheter ablation.

AF screening

Direct-to-consumer technologies such as handheld or wearable devices, and smartphone apps have provided new screening tools for AF^8,25,26. Clinical use of these screening options requires them to be clinically validated, as well as proof that they do improve outcomes in the general population if used more widely than for targeted risk factor-based screening²⁷.

The SAFE study showed that pulse palpation in older patients in a routine or opportunistic fashion increased overall AF detection²⁸. However, this study did not have a control group. The STROKESTOP study targeted people in Sweden who were 75–76 years old. Nearly 29,000 individuals were randomly assigned to screening versus control groups, with the former using a handheld ECG transmitter to record intermittent ECGs for 14 days²⁹. Notably, 51.3% of people who were invited underwent screening, of whom 3% were diagnosed with AF, and subsequently started on anticoagulation. The combined endpoint was a composite of ischemic or hemorrhagic stroke, systemic embolism, bleeding leading to hospitalization, and all-cause death. The screening led to a small but significant reduction in the combined endpoint of ischemic or hemorrhagic stroke, systemic embolism, bleeding leading to hospitalization, and all-cause death.

Several studies using different types of smartwatches have been conducted. The Apple Heart, Huawei Heart, and Fitbit studies employed smartwatch PPG technology for the detection of irregular pulses, followed by the use of ECG patch monitoring to confirm any suspected AF diagnoses^30,31,32.

In the Apple Heart Study, which encompassed nearly 420,000 participants, an irregular pulse notification rate of 0.5% was observed, with 34% of these individuals with irregular pulse notifications ultimately being diagnosed with AF through ECG. Similarly, the Huawei Heart Study and the Fitbit study demonstrated large-scale smartwatches enabled monitoring and detection of AF. However, all these studies lacked randomization and conventional control groups as part of their study design.

In the eBRAVE-AF randomized cross-over clinical trial, over 5500 individuals without AF were randomly assigned to digital screening using a smartphone app or usual care. A certified app was used for digital screening to monitor pulse waves, with abnormal results confirmed by external ECG loop recorders. The primary goal was to identify newly diagnosed AF within 6 months and treat it with oral anticoagulation. The trial found that digital screening more than doubled the detection rate of treatment-relevant AF compared to usual care, with odds ratios of 2.12 and 2.75 in the two phases of the trial³³.

The ongoing HEARTLINE trial (a heart health study using digital technology to investigate if early AF diagnosis reduces the risk of thromboembolic events like stroke in the real-world environment) is a randomized controlled research study that aims to assess if accessible technologies such as Apple Watch with a heart health engagement program, can help with early detection of AF and potentially improve clinical outcomes³⁴. The LOOP trial evaluated whether AF screening-based anticoagulant use can prevent stroke in high-risk individuals. A total of 6004 patients (25% implantable loop recorder (ILR) monitoring and 75% usual care) were followed for 64.5 months. AF was diagnosed in 31% of the ILR group and 12% of the control group. Although ILR screening resulted in a three-times increase in AF detection and anticoagulation initiation, no significant reduction in the risk of stroke or systemic arterial embolism was noted. These findings might suggest that not all screen-detected AF merits anticoagulation^29,35. Handheld ECG monitors such as MyDiagnostick or Merlin have also been studied; while these rely on automated algorithms and have sensitivities varying between 93% and 100%, detection rates of new AF have ranged from 0.9% to 7.4% only²⁶.

While the aforementioned studies may suggest a potential role for population-wide AF screening, the modality of screening (short vs. long-term) and the characteristics of the screened population (including age) impacted the ability to use the tool and any subsequent detection of AF. The mean age of participants in the Huawei Study was 35 years, however, there was a mean age of 54 years in the suspected AF group³¹. In the Apple Watch study, only 5.9% of the total cohort were over 65 years of age³⁰. These results suggest that smartphone/app/watch-based screening is likely to be of most value in older people given the high prevalence of AF in this population. However, lower technology literacy among older patients may lead to under-utilization of these tools leading to the continued presence of undiagnosed AF.

Rhythm tracings and PPG-enabled technologies do not confirm the diagnosis of AF, and subsequent ECG or continuous ECG recording is generally required for diagnosis. The variability in sensitivity and specificity of these monitoring modalities shown in the above trials raises concerns about false positive diagnoses leading to additional unnecessary tests, resulting in, psychological stress and additional costs²⁷.

The 2020 ESC guidelines on AF screening have advised caution regarding the routine use of screening tools other than the standard of care ECG, especially given that current data on these tools were generated in observational cohorts, and no head-to-head comparisons have been reported²⁷. The US Preventive Services Task Force (USPSTF) states that the current evidence is insufficient to assess the balance of benefits and harms of screening for AF³⁶. Intermittent ECG recording ± pulse palpation can contribute to a nearly four-fold increase in new AF detection and provide immediate diagnosis, and it continues to be the gold standard when compared with the aforementioned screening methods²⁷. Opportunistic, as opposed to routine, screening of individuals over the age of 65 years with additional stroke risk factors captured by CHA₂DS₂-VASc score appears to have the best evidence for new AF detection, while providing the best cost-effectiveness³⁷. Analysis of the STROKESTOP study, with a 6.9-year follow-up, demonstrates the cost-effectiveness of population-based AF screening. Screening resulted in gained life years and quality-adjusted life years, with cost-saving demonstrated in 99.2% and 92.7% of simulations, emphasizing its viability, at least in elderly populations³⁸.

It is unclear if treating incidentally diagnosed AF improves patient outcomes in general consumer populations. The ongoing HEARTLINE trial is attempting to answer this question³⁴. In our opinion, at this time these technologies should be used only on an individualized basis, with a caution against overreliance.

AF management

In most individuals, AF is a predictably recurrent disease. Indeed, paroxysmal AF recurrences follow a clustered pattern and persistent AF shows a significant time-dependent pattern of recurrence after rhythm control interventions fail³⁹. Easily accessible, quickly scalable, and user-friendly mHealth technologies offer advantages over conventional tools and make them strong contenders for AF management (Fig. 2).

Mobile technologies can aid the clinicians caring for AF patients in multiple ways including burden assessment, symptom correlation, treatment response monitoring, and drug loading. AF atrial fibrillation.

The clinical utility of these mobile technologies has largely been studied in groups with no prior known AF, directly influencing their predictive values. Validation of accuracy and predictability in those with known AF has not been done for these devices⁴⁰. Indeed, the largest study of mobile ECG and AF, the Apple Heart Study, excluded patients with a reported diagnosis of AF, and its Food and Drug Administration (FDA) clearance states significant limitations during movement and when heart rate is below 50 bpm or above 150 bpm^30,41.

Conceivably, mHealth technologies may help refine stroke risk in AF patients. Scoring systems consider the diagnosis of AF in a binary fashion (present or absent) rather than AF burden. Evidence suggests persistent AF has a higher risk of stroke, HF, and cognitive impairment than paroxysmal AF⁴². Several studies have shown that the risk of stroke is temporally related to the recency of onset of AF^43,44. However, others have shown that ischemic stroke was temporally discordant from AF episodes such that most patients with ischemic stroke did not have AF in the preceding days or weeks⁴⁵. These studies were performed in patients with pacemakers and implantable cardioverter defibrillators (ICDs), who may have higher background rates of ischemic stroke, and have not been reproduced with wearable devices or in lower risk cohorts. Still, quantifying AF burden as a percentage of time in AF over the monitoring period, rather than the absolute number of events, maybe a better indicator of stroke risk and has given rise to the concept of intermittent anticoagulation. This notion was tested in the iCARE-AF, REACT-COM, and TACTIC-AF pilot studies and showed that it is a feasible strategy and may decrease the risk of bleeding in low-risk patients with paroxysmal AF⁴⁶.

Some trials have validated the ability of PPG and ECG-based wearables to detect AF in patients with a prior history of AF, use of anti-arrhythmic drugs, prior cardioversion, and ablation¹⁹. The “pill in the pocket” strategy, which involves carrying medication and using it only when needed, is well-suited for integration with mobile technologies. The potential advantages include a confirmatory validation of symptoms prior to therapy, earlier administration of antiarrhythmic drugs, closer monitoring, and confirmation of therapy success potentially avoiding emergency medical visits and side effects.

Additionally, QT monitoring is feasible with these devices, especially in patients receiving sotalol or dofetilide for rhythm control, as well as monitoring for recurrence after AF ablation⁴⁷. A reliable indicator of recurrence is of the utmost importance as it has been well described that the perception of AF symptoms changes after ablation and recurrence is an AF ablation quality indicator⁴⁸. Obtaining real-time reliable information will help clinicians make informed recommendations when treating patients.

The Heart Rhythm Society and European Heart Rhythm Association publications have offered practical advice concerning the use of wearables by patients for managing cardiovascular health and arrhythmias in various clinical situations. These papers outline tangible pathways for AF screening and management using digital technologies to enhance patient care^49,50,51. Future research efforts will incorporate health apps and AI into treatment protocols as it has already been shown that the use of convolutional neural networks in the detection of AF is feasible⁵².

Role of mHealth technologies in other arrhythmias

SVT

The paroxysmal nature of SVT and its unpredictability makes it a challenging arrhythmia to diagnose, as the arrhythmia often self-terminates before arrival at a healthcare facility for standard ECG recording. The diagnostic yield for traditional monitoring varies from as low as 10% for 24-h Holter monitors to 50–60% for medical-grade event monitors, illustrating that increased duration of monitoring improves yield. mHealth devices (ECG or PPG-based), facilitate relatively inexpensive, long-term rhythm monitoring and may successfully diagnose patients with brief episodes of sustained palpitations^53,54. The resolution of smartphone-based single-lead ECGs is sufficient to differentiate SVT from a common misdiagnosis of sinus tachycardia for experienced ECG readers (89% sensitivity and 91% specificity) and to determine mechanism^55,56. Despite this promise, only 51% of surveyed physicians indicated that they would proceed with an invasive EP study on the basis of a patient-recorded symptomatic, regular tachycardia via a handheld single-lead ECG system, illustrating the need for definitive prospective randomized trials⁵⁷.

PVCs

PVCs are a common cause of palpitations and due to their irregular rhythm, mHealth devices may misdiagnose these as AF. Discrimination algorithms under development may resolve this issue. For example, in PULSE-SMART, a smartphone-based arrhythmia discrimination algorithm reliably discriminated PVCs from sinus rhythm, PACs, and AF with a 96% accuracy^58,59. In another study, a simple computational algorithm filtered and extracted QRS features from an ECG device connected to a smartphone to construct a feature matrix. The algorithm, developed using the MIT-BIH arrhythmia database and clinically validated in a separate cohort of 100 participants, documented a PVC recognition accuracy of 98.69%⁶⁰. While diagnosing and differentiating PVC is improved, quantification of burden may still need to be addressed.

VT

While the evidence supporting the use of smartphones for diagnosing ventricular arrhythmias is limited to case reports, these reports do suggest that this approach could be beneficial. In one case, non-sustained VT correlated with symptoms with exertional pre-syncope, leading to an EP study with induction of sustained right ventricular outflow tract VT and successful ablation⁶¹. In another case, recurrent syncope was associated with a recording of monomorphic VT on Apple Watch leading to a secondary prevention ICD placement⁶².

However, it is important to recognize that the practical application of wearables for VT diagnosis and management in the near future is less likely. While the widespread availability of smartphones allows patients to record SVT, PVCs, and VT, with subsequent accurate physician interpretation; the complexities and potential risks associated with VT necessitate a cautious approach. As VT diagnosis and management often require immediate intervention and specialized medical attention, the incorporation of wearable technology must be carefully evaluated and integrated into a broader clinical context. Clinical trials, such as the multi-center randomized control trial investigation of palpitations in the ED (IPED), may offer guidance on how best to incorporate this rapidly developing technology into clinical practice⁶³.

Role of mHealth in risk-stratification, prevention, and treatment of sudden cardiac death (SCD)

The implementation of mHealth and related technologies could affect the risk assessment and treatment of SCD in three ways: identifying individuals who could benefit from an ICD or other targeted intervention due to their high long-term risk and potential benefits; detecting individuals who may be at imminent risk of cardiac arrest; and reducing the time to defibrillation and/or enhancing the quality of resuscitation to improve the chances of survival⁶⁴ (Fig. 3).

This figure illustrates the various ways in which mHealth technology can be utilized in the prevention and management of sudden cardiac arrest. It showcases the use of ML and AI analysis to identify novel patterns, the application of wearable technologies and bio-sensor data for early detection, and the enhancement of response time and quality of resuscitation through innovative methods such as citizen-responder systems and drone delivery of AEDs. ICD implantable cardioverter defibrillator, CPR cardiopulmonary resuscitation, AED automated external defibrillator.

Identification of individuals at risk of SCD

While ICDs save lives in patients with clinical characteristics known to indicate a high risk of SCD, their impact on decreasing overall sudden death is low, as the large majority of sudden deaths occur in individuals without these indicators^65,66. Better tools for risk prediction, particularly in the general population without known heart disease, are thus critical to decreasing SCA. AI and ML may address this critical need. In one example with immediate clinical relevance, ML has been used to create algorithms that can identify decreased LV function, known to predict ICD benefit, from the standard 12-lead ECG⁶⁷.

Other ML algorithms have used complex computing to identify new variables that may predict the risk of SCA or overall mortality, whether from clinical or ECG patterns of variables^68,69. Many factors predict long-term mortality in the general population. These include measures based on heart rate time series, such as heart rate variability (HRV) or heart rate recovery on stress tests, or nonlinear factors, which probably identify overall physiological stress, transduced through the autonomic nervous system or other processes^70,71. Others identify vulnerable electrophysiological substrates, such as T wave alternans or QRST-integral. There are challenges in moving from prediction to improvement in mortality, for both these prior markers, and newer AI-based markers. To improve mortality, the marker must provide adequate positive and negative predictive value to drive a preventative intervention⁷². Transitioning from predicting mortality to improving it, is challenging for both traditional markers and newer AI-based markers. For a marker to effectively enhance mortality, it must possess sufficient positive and negative predictive values to support a preventive intervention. Alternatively, the marker should pinpoint modifiable physiological processes. While AI holds promise in overcoming these challenges, its effectiveness remains unproven.

Commercial companies advertise that wearable devices measuring HRV can be used as a tool to improve overall health. Limited data in controlled settings showed that HRV monitoring can optimize and avoid overtraining in athletes⁷³. A study of the clinical relevance of diverse parameters captured via a wearable cardioverter-defibrillator (WCD) in recently diagnosed HF patients notably revealed that alterations in heart rate, step count, and HRV observed during WCD usage could be used as prognostic predictors for improvements in left ventricular ejection fraction (LVEF)⁷⁴. However, currently, the role of HRV as a mHealth tool in the general population is untested in systematic clinical trials.

Prediction and identification of imminent cardiac arrest

Wearable watches can identify a fall and connect the wearer to emergency services⁷⁵. Whether sensor-based biometric data, perhaps combined with fall identification and verbal feedback, could provide identification of an imminent SCA, is intriguing. In one early Holter-based study, dynamic changes in repolarization were seen in a 10-min time frame preceding ventricular arrhythmia⁷⁶. The concept of a wearable device that can truly provide a warning signal of impending arrest is exciting, but far in the future⁷⁷.

Improving the time to resuscitation

Improving the time to the performance of cardiopulmonary resuscitation (CPR), and defibrillation, is the most important, and modifiable, contributor to survival from SCA, and is perhaps the most important use of mobile technology for SCA to date. Since its first report in 2007, citizen-responder systems have not only increased layperson CPR, but also have improved survival to hospital discharge by 50%^78,79. In these systems, citizens willing to provide assistance register in a first-responder network and those closest to a SCA patient based on geolocation are notified by text. Similarly, the use of mobile apps to identify the nearest automatic external defibrillator decreases the time to access the AED and may improve outcomes⁸⁰. ML may further facilitate the use of drones for faster delivery of AEDs, although data on these systems to date are based only on simulations⁸¹. In addition to decreasing the time for CPR and defibrillation, smartphone apps may also improve the quality of resuscitation. In one simulation study, smartphone video-based analysis of CPR performance improved the quality of compressions⁸². The American Heart Association has recognized the importance of further research in leveraging digital strategies to improve survival from cardiac arrest^83,84.

Role of mobile technologies in arrhythmia care in children

While mHealth technologies are developed and validated in the adult population, there is rapid uptake of these tools in the pediatric and congenital heart populations. Studies show 53% of children have a tablet by age 11 and 84% own a smartphone at age 13^85,86. Though the use of wearables in this group is poorly defined, their technological savvy may enable quick adoption.

A few mHealth technologies have been tested and validated compared to standard 12-lead ECGs in children, including the Alive Cor Kardia Monitor and Apple Watch for symptom-rhythm correlation and QT measurement with good patient/provider usability and data quality^55,87,88,89. Automated algorithmic interpretation has been limited often requiring physicians over-read to ensure accuracy^87,88. This should not be surprising given the training datasets used for algorithm development are from older adults with slower baseline heart rates (HRs) and tachycardia rates in contrast to the higher resting baseline HRs and tachycardia rates seen in younger patients. Patch-based wearable devices were more user-friendly than Holter and often provided high-quality data but may be limited by small chest surface area and patient intolerance to adhesive⁹⁰.

Use of mHealth application for care of non-arrhythmia conditions

mHealth applications have improved the care of patients with other cardiac conditions, which may have direct implications on atrial and ventricular arrhythmia development and disease course in these patients. We therefore highlight a few advances below.

Coronary artery disease

Various smart device-based mobile ECG systems have been studied for earlier detection of myocardial infarction (MI). In a study of multichannel ECG using a smartwatch, the sensitivity, and specificity for the diagnosis of ST elevation MI and non-ST elevation MI was greater than 90%⁹¹. Similarly, an FDA-approved ICM with an IS-1 lead implanted into the right ventricular apex for real-time monitoring of ST segment changes, allowed for early detection with 55% of patients presenting within 2 h of the MI⁹². Apart from early detection of MI, mHealth applications using smartphones and fitness trackers help improve adherence and cost-effectiveness of cardiac rehabilitation and also promote risk factor modification and compliance^93,94. Larger studies are needed to validate the use of mHealth applications in both the early detection of MI and in increasing cardiac rehabilitation participation.

HF

mHealth could have a significant role in preventing HF hospitalization by recognizing HF exacerbations earlier based on vital signs and symptoms, thoracic impedance, and hemodynamic monitoring systems. In a meta-analysis of studies on mHealth interventions, there was a significant reduction in all-cause and CV mortality, as well as HF hospitalization⁹⁵. Thoracic impedance monitoring with CRT and ICD devices improves HF management. The remote dielectric sensing system detects fluid overload in the lungs and thereby provides objective measurement of pulmonary edema leading to early therapy, preventing HF hospitalization⁹⁶. Algorithms such as Heartlogic® identify early HF exacerbations by monitoring multiple parameters such as heart sounds, rate, and activity plus thoracic impedance, respiratory rate, and tidal volume. ALLEVIATE-HF is an ongoing study utilizing an ICM-based algorithm to reduce HF events. Hemodynamic monitoring with a PA pressure monitor (eg. CardioMEMS) has shown a significant reduction in HF hospitalizations by 37% over 15 months. Supplementary Table 1 outlines the relevant clinical trials on mHealth in HF.

Hypertension

Multiple studies have shown that mHealth platforms that allow tracking of BP by clinicians helped lower systolic and diastolic BP⁹⁷. Currently, research studies are underway to show the feasibility and validity of PPG-based home BP assessment through wearables for continuous ambulatory checks leading to improved personalized hypertension management⁹⁸.

DM

DM was one of the first diseases where mHealth-based management was successfully explored. Various mHealth platforms that encourage self-monitoring of blood glucose monitoring, medication, and diet compliance have been successful (0.3–0.4% reduction in HbA1C) enough to be recommended by international societies. Continuous blood glucose monitoring, and insulin delivery through smart pens and automated pumps have positively impacted DM management⁹⁹. A full discussion of the transformative roles of mHealth and digital health technologies in diabetes management is beyond the scope of this article, but a comprehensive review on the subject has recently been published¹⁰⁰.

Sleep apnea

Various mHealth applications have been used for diagnosis and follow-up of sleep apnea (OSA). Sleep trackers such as smart watches, wearable rings, smart mattresses, and other smartphone-based sensors can be used to screen for OSA. Similarly, newer CPAP machines allow better integration with mHealth allowing for tracking of CPAP use and helping improve compliance¹⁰¹.

mHealth applications in clinical trials

mHealth devices are revolutionizing clinical trials by enabling a site-less approach, where participants can join and provide data remotely, without needing to visit a trial site¹⁰². These devices use Internet of Things (IoT) technology to actively and passively collect a broad range of health data. This method not only reduces the need for in-person visits but also expands trial participation to include people who are typically underrepresented, such as those with mobility-limiting disabilities or those living in remote areas with limited access to research centers^102,103. Outside of cardiology, mHealth devices using frameworks such as Apple’s open-source Research Kit ( have utilized smart device sensor data to study a variety of topics including increasing physical activity, rheumatoid arthritis symptom management, Parkinson’s disease progression, seizure detection in epilepsy, and geriatric fall prevention^{104,105,106,107,108}.

Before the advent of mHealth technologies, ambulatory ECG monitoring had already been widely utilized as a data collection modality in clinical trials. For example, the Catheter Ablation vs Antiarrhythmic Drug Therapy for AF (CABANA) trial employed a proprietary telephonic transmission two-lead ECG mobile cardiac telemetry device that facilitated patient-activated monitoring^109,110. The Apple Heart Trial and Huawei Heart Study demonstrated the feasibility of PPG smart watch-based screening for AF^30,31.

Traditional rhythm monitoring tools employed as gold standards in clinical trials are limited by the need for resource-intensive mobile cardiac telemetry and the time lag between data collection and interpretation in the case of patch monitors. These limitations can be overcome by mHealth for QT interval monitoring, which is an important consideration in pharmaceutical trials. Many potential therapeutics could prolong the QT interval, potentially resulting in increased risk for torsades de pointes (TdP)¹¹¹. These trials require real-time monitoring so that investigators can assess the risk of QT prolongation and intervene in a timely manner to prevent TdP. A recent remote randomized trial of the known QT-prolonging agent’s hydroxychloroquine and azithromycin for the treatment of COVID-19 utilized a six-lead KardiaMobile 6 L for ECG collection, allowing for accurate and timely ascertainment of the QTc, core laboratory validation and notification of results to the patient and treating physician^102,112.

There are several potential advantages to using mHealth in cardiovascular research in general, and cardiac arrhythmia research in particular (Fig. 4). By digitizing trial elements, including enrollment, clinical history gathering, follow-up visits, and outcome assessment, the cost per participant may be reduced. In addition, this may allow broader representation by including participants who have travel impediments or are historically underrepresented in trials due to structural disparities³⁰. While this approach has the potential to enhance the diversity and inclusivity of participants, it comes with an inherent risk of disadvantaging those with limited means and lower technological literacy¹¹³. Careful attention must be paid to diversity and equity as the adoption of mHealth tools for clinical research continues to grow²⁶. Further, as highlighted by a 50% attrition rate in completion of the Apple Heart virtual study visit, future studies relying on mHealth will need to account for potentially increased loss to follow-up³⁰.

Integrating mobile technologies in clinical trials offers a range of benefits, from enabling frequent monitoring to increasing access for remote populations. These technologies facilitate quick data transfer, enhance patient safety through rapid alert notifications, and reduce costs associated with transportation and device expenses. Additionally, centralized data monitoring ensures uniform quality control and supports the globalization of clinical trials.