AFib Detection Research Articles

Smartphone motion sensor based detection of atrial fibrillation in populations referred to cardiology

Abstract Introduction The motion sensors of a common smartphone when placed on chest can be used to detect vibrations caused by cardiac motion. The concept has been shown to detect atrial fibrillation (AFib) in clinical setting with 96% accuracy. Purpose We sought to examine how the smartphone motion sensor based AFib detection performs in various populations referred to cardiology pooled from 3 different studies addressing arrhythmias, heart failure or suspected coronary artery disease. Methods From a total of 1 524 patient recordings 748 recordings were from men (49%). The mean age of the patients was 64 years (range 20-94 years) and mean BMI was 28.9kg/m2 (range 15.2-63.7kg/m2). The signals were collected by placing a smartphone on chest of the patients in supine position and then analyzed with the CardioSignal solution. From the recordings two 60 second strips were analyzed. A simultaneous 5-lead or 1-lead ECG data was collected for reference. The ECG signals were analyzed for quality and rhythm diagnostics by 2 physicians blinded for the motion sensor AFib detection result. Finally, a head-to-head comparison was conducted for the motion sensor AFib detection result and the ECG data. Results From the 1524 patient recordings the final analysis set included 237 individuals with AFib, whereas the rest had sinus rhythm. The sensitivity to detect AFib was 91.9% and specificity 98.7%. The positive predictive value was 91.0% (88.3% to 94.2%) and the negative predictive value 98.3% (95.2% to 99.1%). The accuracy was 97.0% (94.7% to 97.7%). Conclusions A smartphone-based assessment of AFib using the embedded motion sensors seem to yield very good detection results, especially specificity. The accuracy of the results is comparable to available ECG and PPG based AFib detection solutions.

Validation of an artificial intelligence system for holter ECG analysis

Abstract Introduction Holter ECG is a commonly used tool to identify heart rhythm disorders. As physician-based analysis of such Holter ECG is time consuming, various approaches using computerized interpretation have been developed. However, accuracy of these tools remains inferior to physician-based analysis. Thus, this study assesses the diagnostic performance of an artificial intelligence (AI)-powered Holter ECG service in comparison to results from reference databases as well as with physician-based evaluation. Methods 231 ECG recordings, covering 2,261 hours of ECG with annotations for 10,024,315 heartbeats from the following reference databases were used: MIT-BIH Arrhythmia (MITDB), American Heart Association ECG (AHDB), MIT-BIH Atrial Fibrillation (AFDB), Long Term Atrial Fibrillation (LTAFDB). Those recordings were originally collected from a mixed population of inpatients and outpatients between 1975 and 2008. Analysis of these recordings regarding the detection of QRS complexes, QRS classification and detection of Atrial Fibrillation (AF) was performed using a novel service for AI-based ECG analysis. Results were compared to the diagnosis stated within the database. QRS complex detection accuracy was performed on recordings from MITDB and AHADB by checking if time positions are within 150 ms to reference annotations. The QRS classification in three classes (normal, supraventricular, ventricular) was then compared for each detected heartbeat. Mismatches were adjudicated manually by three experts. Episodes of Atrial Fibrillation were detected in recordings from MITDB, AFDB and LTAFDB and compared to the reference episodes from the databases. The accuracy was evaluated based on the overlap with the reference annotations in the time-domain. Results Sensitivity and Precision of heartbeat detection compared to the reference databases exceeded 99% Figure 1 displays the results of heartbeat classification. Normal heartbeats were classified with a sensitivity of 99.6%, precision of 99.9% and specificity of 99.6%. For ventricular heartbeats, a sensitivity of 99.58 %, precision of 99.88 and specificity of 99.99% were achieved. Premature supraventricular contractions were recognized with a sensitivity of 99.58 %, precision of 100% and specificity of 100%. When comparing the 248,187 QRS complex classifications of MITDB and AHADB to the AI-powered analysis, 128 discrepancies were revealed. 43 were due to a discrepancy of convention as these beats were in sections that were also marked as AF episodes by the databases and the AI-powered analysis marks all beats in AF episodes as normal. Of the 85 remaining discrepancies, more than 87% were adjudicated in favor of the AI-powered analysis by the experts. Figure 2 shows the comparison of AF detection results showing Sensitivity and Precision above 99%. Conclusion AI-powered Holter ECG analysis compared to reference ECG stripes exceed 99% of sensitivity, specificity and precision.QRS complex detectionAF/ AFib detection

Smart watch applications in atrial fibrillation detection: Current state and future directions.

Atrial fibrillation (Afib) is a prevalent chronic arrhythmia associated with severe complications, including stroke, heart failure, and increased mortality. This review explores the use of smartwatches for Afib detection, addressing the limitations of current monitoring methods and emphasizing the potential of wearable technology in revolutionizing healthcare. Current Afib detection methods, such as electrocardiography, have limitations in sensitivity and specificity. Smartwatches with advanced sensors offer continuous monitoring, improving the chances of detecting asymptomatic and paroxysmal Afib. The review meticulously examines major clinical trials studying Afib detection using smartwatches, including the landmark Apple Heart Study and ongoing trials such as the Heart Watch, Heartline, and Fitbit Heart Study. Detailed summaries of participant numbers, smartwatch devices used, and key findings are presented. It also comments on the cost-effectiveness and scalability of smartwatch-based screening, highlighting the potential to reduce healthcare costs and improve patient outcomes. The integration of wearable technology into healthcare can lead to earlier diagnosis, improved patient engagement, and enhanced cardiac health monitoring. Despite ethical considerations and disparities, the potential benefits outweigh the challenges. This review calls for increased awareness, collaboration with insurance companies, and ongoing research efforts to optimize smartwatch accuracy and encourage widespread adoption of Afib detection. With insights from major trials, this review serves as a comprehensive reference for healthcare professionals and policymakers, guiding future strategies in the early diagnosis and management of atrial fibrillation.

Evaluation of the performance of a machine learning based atrial fibrillation screening algorithm using an oscillometric blood pressure monitor

Blood pressure monitors (BPMs) with atrial fibrillation (AFib) detection function can be used to detect AFib early. However, conventional algorithms require multiple BP measurements. Here, the feasibility of a machine-learning-based approach for AFib detection through single BP measurement was evaluated. First, a custom AdaBoost-based software, which analyzes the pulse-to-pulse interval (PPI) pattern and classifies it based on AFib detection, was created. Then, its classification performance was validated. For the validation study, PPI and standard 12-lead electrocardiogram (ECG) datasets were collected from 79 and 92 Japanese participants with and without AFib, respectively. PPI data were obtained using two different BPMs. All ECG results were interpreted by cardiologists. The custom software output for the PPI dataset and ECG interpreted results was compared, and the sensitivity and specificity were calculated. A sensitivity and specificity for PPI from main device were 97.5% (95% confidence interval [CI] 91.2–99.3%) and 98.9 (95% CI 94.1–99.8), respectively. No significant differences in sensitivity and specificity were observed in the subgroup analysis between different devices, age groups, and arm size groups. These results reflect the high accuracy and robustness of this AFib algorithm using a single BP measurement and supports its use for widespread AFib screening.

A Crowdsourced AI Framework for Atrial Fibrillation Detection in Apple Watch and Kardia Mobile ECGs.

Background: Atrial fibrillation (AFib) detection via mobile ECG devices is promising, but algorithms often struggle to generalize across diverse datasets and platforms, limiting their real-world applicability. Objective: This study aims to develop a robust, generalizable AFib detection approach for mobile ECG devices using crowdsourced algorithms. Methods: We developed a voting algorithm using random forest, integrating six open-source AFib detection algorithms from the PhysioNet Challenge. The algorithm was trained on an AliveCor dataset and tested on two disjoint AliveCor datasets and one Apple Watch dataset. Results: The voting algorithm outperformed the base algorithms across all metrics: the average of sensitivity (0.884), specificity (0.988), PPV (0.917), NPV (0.985), and F1-score (0.943) on all datasets. It also demonstrated the least variability among datasets, signifying its highest robustness and effectiveness in diverse data environments. Moreover, it surpassed Apple's algorithm on all metrics and showed higher specificity but lower sensitivity than AliveCor's Kardia algorithm. Conclusions: This study demonstrates the potential of crowdsourced, multi-algorithmic strategies in enhancing AFib detection. Our approach shows robust cross-platform performance, addressing key generalization challenges in AI-enabled cardiac monitoring and underlining the potential for collaborative algorithms in wearable monitoring devices.

The potentially underestimated influence of atrial fibrillation on NT-proBNP in chest pain

Abstract Introduction/Purpose NT-proBNP is a widely used diagnostic marker in the emergency department (ED). The main objective of the current study was to determine the association of NT-proBNP in patients with chest pain presenting with sinus rhythm (SR) compared to atrial fibrillation (AFib). Hereby, a focus was the evaluation of serum NT-proBNP head-to-head to urinary NT-proBNP. Methods In this study, urinary as well as blood concentrations of NT-proBNP were measured in a total of 601 patients, who presented with acute chest pain in the ED (mean age = 61.1 y, SD = 16.9 y). In the initial ECG 68 patients suffered from AFib and 533 showed SR. The participants were categorized according to their diagnosis into acute coronary syndrome (ACS, AFib n = 28, SR n = 261), acute decompensated heart failure (ADHF, AFib n= 16, SR n = 21), non-ischemic chest pain without congestion (NICP, AFib n = 24, SR n = 96) or chest-wall syndrome (CWS, SR n = 155). Results In the overall study cohort, patients with AFib showed significantly higher concentrations of urinary as well as blood NT-proBNP compared to those in sinus rhythm (each p < 0.001). Additionally, among patients with acute coronary syndrome (ACS), there were significantly increased biomarker concentrations in patients presenting with AFib compared to sinus rhythm (each p < 0.001, AUCblood = 0.801, AUCurine = 0.701), similarly for patients categorized into NICP (each p < 0.001, AUCblood = 0.833, AUCurine = 0.756). The same pattern was observed among the ACS subcategories (each p < 0.001). Regarding ROC analysis, promising predictive values were shown for detection of AFib: STEMI (AUCblood = 0.902, AUCurine = 0.867), NSTEMI (AUCblood = 0.757, AUCurine = 0.696) and UA (AUCblood = 0.821, AUCurine = 0.681). Comparing ROC curves, significant differences were noted in UA (p < 0.001), opposite to STEMI as well as NSTEMI (each p = n. s.). Regarding ADHF, paradoxically higher NT-proBNP concentrations were noted among patients showing sinus rhythms compared to AFib. Furthermore, a multivariate binary regression analysis showed blood NT-proBNP and age as independent predictors of atrial fibrillation (p < 0.05), opposite to urinary NT-proBNP, high-sensitivity troponin T and diabetes (each p = n. s.). Conclusion Urinary and to a larger extent blood NT-proBNP showed significantly higher concentrations in AFib. In conclusion, in ED patients elevated NT-proBNP may potentially be frequently caused by a coexisting AFib. Regarding ADHF patients, the effect of AFib might be consumed by other conditions that have a greater impact on NT-proBNP.

Evaluation of the Diagnostic Performance and Efficacy of Wearable Electrocardiogram Monitoring for Arrhythmia Detection after Cardiac Surgery.

Postoperative atrial fibrillation (A-fib) is a serious complication of cardiac surgery that is associated with increased mortality and morbidity. Traditional 24-hour Holter monitors have limitations, which have prompted the development of innovative wearable electrocardiogram (ECG) monitoring devices. This study assessed a patch-type wearable ECG device (MobiCARE-MC100) for monitoring A-fib in patients undergoing cardiac surgery and compared it with 24-hour Holter ECG monitoring. This was a single-center, prospective, investigator-initiated cohort study that included 39 patients who underwent cardiac surgery between July 2021 and June 2022. Patients underwent simultaneous monitoring with both conventional Holter and patchtype ECG devices for 24 hours. The Holter device was then removed, and patch-type monitoring continued for an additional 48 hours, to determine whether extended monitoring provided benefits in the detection of A-fib. This 72-hour ECG monitoring study included 39 patients, with an average age of 62.2 years, comprising 29 men (74.4%) and 10 women (25.6%). In the initial 24 hours, both monitoring techniques identified the same number of paroxysmal A-fib in 7 out of 39 patients. After 24 hours of monitoring, during the additional 48-hour assessment using the patch-type ECG device, an increase in A-fib burden (9%→38%) was observed in 1 patient. Most patients reported no significant discomfort while using the MobiCARE device. In patients who underwent cardiac surgery, the mobiCARE device demonstrated diagnostic accuracy comparable to that of the conventional Holter monitoring system.

Association of device detected atrial high rate episodes with embolic events

Abstract Background Subclinical atrial fibrillation (AFib) is a common etiology of embolic events such as embolic strokes of undetermined source. With the advent of cardiac implantable electronic devices (CIED), there has been increased detection of subclinical AFib, or atrial high rate episodes (AHRE) in asymptomatic patients without a known diagnosis of AFib. The clinical significance of AHRE detected with CIED remains poorly characterized with studies reporting conflicting results. The purpose of this meta-analysis is to evaluate the association of AHRE with cerebrovascular accidents (CVA), transient ischemic attacks (TIA), and systemic embolic events. Methods A literature search was conducted for studies reporting on the association of subclinical AFib or AHRE with CVA, TIA, and systemic embolic events. The search included the following databases: Ovid MEDLINE, EMBASE, Google Scholar, and Web of Science. The search was not restricted to time or publication status. Results A total of 19 studies and 50,968 patients (12,879 with AHRE vs 38,089 without AHRE) were included. Mean follow-up was 32 months, mean age was 72 years old, 53.3% were male. Patients with device detected AHRE had increased risk of composite endpoint of CVA, TIA, and systemic embolic events (OR 2.98, 95% CI 2.03-4.39; P<0.01). Subgroup analysis by maximum AHRE burden demonstrated that both maximum AHRE burden >24 hours and burden <24 hours are associated with increased risk of CVA, TIA, and systemic embolic events (OR 4.82, 95% CI 2.07-11.24; p<0.01; OR 1.55, 95% CI 1.11-2.15; p<0.01). The overall effect size is larger for maximum AHRE burden >24 hours compared to <24 hours (Z=3.64 vs Z=2.61). Conclusions Our results suggest that AHRE detected with CIED is associated with increased risk of CVA, TIA, and systemic embolic events regardless of maximum AHRE burden >24 hours or <24 hours. Additional studies are required to determine the utility anticoagulation therapy for treatment of asymptomatic patients with device detected AHRE.

Solutions for Atrial Fibrillation Edvocacy (SAFE): Improving Awareness And Access To Afib Screening For Detection And Referral For Treatment.

Background: Atrial Fibrillation (Afib) can lead to stroke and heart failure, and early detection of Afib is an effective method of preventing these life-threatening conditions. An estimated 2.7 million Americans are living with Afib1, a number that is expected to rise dramatically in the coming years. Methods: The aim of this demonstration project was to create an additional access point in the community at local pharmacies for Afib screening, detection, and referral to physicians for follow-up and initiation of evidence-based therapy when appropriate. This prospective research study was conducted with 14 community pharmacies across the US, in which a total of 650 patients were screened for Afib. Pharmacists conducted SAFEty Risk Assessments that consisted of completion of a Stroke Risk Scorecard and EKG determination utilizing AliveCor's KardiaMobile® 6L device. Results: In 552 (82.5%) of 669 total EKG readings, a "normal" rhythm was detected, and in 117 (17.5%) EKG readings an abnormal detection occurred. A total of 12 out of 650 patients (1.8%) received EKG readings of Afib, which is greater than double the expected prevalence of Afib in the US (0.81%), a statistically significant finding (p < 0.0001). Other notable findings included 42 (6.3%) EKG readings of Wide QRS, and 26 (3.9%) EKG readings of tachycardia. A total of 44 patients were referred to physicians for follow-up on their risk for Afib. Conclusions: Community pharmacies offer a unique, valuable access point for patients to receive Afib screenings. Pharmacists are well positioned to make a significant contribution in the cardiovascular health of their patients and increase the value of team-based health care.

A Study of R-R Interval Transition Matrix Features for Machine Learning Algorithms in AFib Detection

Atrial Fibrillation (AFib) is a heart condition that occurs when electrophysiological malformations within heart tissues cause the atria to lose coordination with the ventricles, resulting in "irregularly irregular" heartbeats. Because symptoms are subtle and unpredictable, AFib diagnosis is often difficult or delayed. One possible solution is to build a system which predicts AFib based on the variability of R-R intervals (the distances between two R-peaks). This research aims to incorporate the transition matrix as a novel measure of R-R variability, while combining three segmentation schemes and two feature importance measures to systematically analyze the significance of individual features. The MIT-BIH dataset was first divided into three segmentation schemes, consisting of 5-s, 10-s, and 25-s subsets. In total, 21 various features, including the transition matrix features, were extracted from these subsets and used for the training of 11 machine learning classifiers. Next, permutation importance and tree-based feature importance calculations determined the most predictive features for each model. In summary, with Leave-One-Person-Out Cross Validation, classifiers under the 25-s segmentation scheme produced the best accuracies; specifically, Gradient Boosting (96.08%), Light Gradient Boosting (96.11%), and Extreme Gradient Boosting (96.30%). Among eleven classifiers, the three gradient boosting models and Random Forest exhibited the highest overall performance across all segmentation schemes. Moreover, the permutation and tree-based importance results demonstrated that the transition matrix features were most significant with longer subset lengths.

Morphological Autoencoders for Beat-by-Beat Atrial Fibrillation Detection Using Single-Lead ECG

Engineered feature extraction can compromise the ability of Atrial Fibrillation (AFib) detection algorithms to deliver near real-time results. Autoencoders (AEs) can be used as an automatic feature extraction tool, tailoring the resulting features to a specific classification task. By coupling an encoder to a classifier, it is possible to reduce the dimension of the Electrocardiogram (ECG) heartbeat waveforms and classify them. In this work we show that morphological features extracted using a Sparse AE are sufficient to distinguish AFib from Normal Sinus Rhythm (NSR) beats. In addition to the morphological features, rhythm information was included in the model using a proposed short-term feature called Local Change of Successive Differences (LCSD). Using single-lead ECG recordings from two referenced public databases, and with features from the AE, the model was able to achieve an F1-score of 88.8%. These results show that morphological features appear to be a distinct and sufficient factor for detecting AFib in ECG recordings, especially when designed for patient-specific applications. This is an advantage over state-of-the-art algorithms that need longer acquisition times to extract engineered rhythm features, which also requires careful preprocessing steps. To the best of our knowledge, this is the first work that presents a near real-time morphological approach for AFib detection under naturalistic ECG acquisition with a mobile device.

A Study of R-R Interval Transition Matrix Features for Machine Learning Algorithms in AFib Detection

A Study of R-R Interval Transition Matrix Features for Machine Learning Algorithms in AFib Detection

Detecting Atrial Fibrillation in Real Time Based on PPG via Two CNNs for Quality Assessment and Detection

Real-time detection of atrial fibrillation (AFib) is made possible by the quality assessment via a 1-D convolutional neural network (1D-CNN) in a processor of a photoplethysmography (PPG) sensor patch and a 2-D convolutional neural network (2D-CNN) for AFib detection in cloud. The 1D-CNN is able to filter out the unqualified PPG that is contaminated by motion artifacts and/or ambient light interference. The remaining qualified PPG is then inputted to another built 2D-CNN for detecting AFib. This 1D-CNN consists of four layers of convolutions and max pooling, one long short-term memory (LSTM), and an output dense layer. The 2D-CNN is pretrained based on the electrocardiography (ECG) data from multiparameter intelligent monitoring in intensive care (MIMIC) III database, for which the RR-intervals (RRIs) of ECG data are first extracted in Poincaré images and then regarded as input features to the model for training. This 2D-CNN has also four layers of convolutions and max pooling and four output dense layers. The pretrained model is next fine-tuned based on peak-to-peak intervals (PPIs) of PPG measured from wearable devices as input features for detecting AFib effectively. The quality-assessment 1D-CNN model is implemented in the wearable device to transmit only qualified data to the 2D-CNN model in cloud for AFib detection, achieving power efficiency. Both models are trained by the Adam optimizer. To validate the models, the PPIs of PPG were collected to evaluate the performance of the established models in real time. Experimental results show that the fine-tuned 2D-CNN for AFib detection achieves the accuracy, sensitivity, and specificity were 98.08%, 96.82%, and 98.86%, respectively, the most favorable as opposed to other reported works based on PPG. The models are able to not only assist clinicians in AFib detection but also provide a mechanism to detect AFib via wearable devices in real time.

Use of digital health applications for the detection of atrial fibrillation.

The advances in health care technologies over the last decade have led to improved capabilities in the use of digital health applications (DiHA) for the detection of atrial fibrillation (AFib). Thus, home-based remote heart rhythm monitoring is facilitated by smartphones or smartwatches alone or combined with external sensors. The available products differ in terms of type of application (wearable vs. handheld) and the technique used for rhythm detection (electrocardiography [ECG] vs. photoplethysmography [PPG]). While ECG-based algorithms often require additional sensors, PPG utilizes techniques integrated in smartphones or smartwatches. Algorithms based on artificial intelligence allow for the automated diagnosis of AFib, enabling high diagnostic accuracy for both ECG-based and PPG-based DiHA. Advantages for clinical use result from the widespread accessibility of rhythm monitoring, thereby permitting earlier diagnosis and higher AFib detection rates. DiHA are also useful for the follow-up of patients with known AFib by monitoring the success of therapeutic interventions to restore sinus rhythm, e.g. catheter ablation. Although some studies strongly suggest apotential benefit for the use of DiHA in the setting of AFib, the overall evidence for an improvement in hard, clinical endpoints and positive effects on clinical care is scarce. To enhance the acceptance of DiHA use in daily practice, more studies evaluating their clinical benefits for the detection of AFib are required. Moreover, most of the applications are still not reimbursable, although the German Digital Health Care Act (Digitale-Versorgung-Gesetz, DVG) made reimbursement possible in principle in 2019.

Boosting the performance of a fusion-based lead-specific a-fib detection algorithm by voting between multiple leads of independent algorithms

Boosting the performance of a fusion-based lead-specific a-fib detection algorithm by voting between multiple leads of independent algorithms

Visual Reassessment with Flux-Interval Plot Configuration after Automatic Classification for Accurate Atrial Fibrillation Detection by Photoplethysmography.

Atrial fibrillation (AFib) is a common type of arrhythmia that is often clinically asymptomatic, which increases the risk of stroke significantly but can be prevented with anticoagulation. The photoplethysmogram (PPG) has recently attracted a lot of attention as a surrogate for electrocardiography (ECG) on atrial fibrillation (AFib) detection, with its out-of-hospital usability for rapid screening or long-term monitoring. Previous studies on AFib detection via PPG signals have achieved good results, but were short of intuitive criteria like ECG p-wave absence or not, especially while using interval randomness to detect AFib suffering from conjunction with premature contractions (PAC/PVC). In this study, we newly developed a PPG flux (pulse amplitude) and interval plots-based methodology, simply comprising an irregularity index threshold of 20 and regression error threshold of 0.06 for the precise automatic detection of AFib. The proposed method with automated detection on AFib shows a combined sensitivity, specificity, accuracy, and precision of 1, 0.995, 0.995, and 0.952 across the 460 samples. Furthermore, the flux-interval plot configuration also acts as a very intuitive tool for visual reassessment to confirm the automatic detection of AFib by its distinctive plot pattern compared to other cardiac rhythms. The study demonstrated that exclusive 2 false-positive cases could be corrected after the reassessment. With the methodology’s background theory well established, the detection process automated and visualized, and the PPG sensors already extensively used, this technology is very user-friendly and convincing for promoted to in-house AFib diagnostics.

Direct-to-consumer detection of atrial fibrillation in a smartwatch electrocardiogram: Medical overuse, medicalisation and the experience of consumers

The Apple Watch Electrocardiogram (ECG) is a digital feature that detects signs of atrial fibrillation (AFib), a heart arrhythmia that can lead to stroke. Unlike AFib detection offered in a clinical setting to patients or those at higher risk, the Apple Watch ECG is a direct-to-consumer (DTC) product marketed to the healthy as a self-performed medical test. The feature is incorporated in the design as one of many applications in a multifunctional smartwatch. The Apple Watch ECG signals the movement of consumer wearables into the domain of medical devices, and the highly contested practice of AFib screening. This article examines how this technology produces new avenues for medical overuse among people who are unlikely to gain clinical benefit, and who as active consumers become medicalised via taking part in disease-specific monitoring. Interviews with Apple Watch ECG consumers suggest their strongly trust in the Apple brand is amplified by the promissory quality of screening and technical innovation. What would otherwise be a costly clinical procedure is condensed into an accessible function in a wearable device. Consequently, AFib screening loses its clinical purpose, and is instead refashioned into a component of healthy lifestyle behaviour. Repeat self-screening becomes ‘bundled’ with similar health monitoring practices routinised in consumer wearables. Active uptake supports a market for the product and adds to the reputation of the Apple brand such that consumers become complicit in legitimising a medical practice that has limited clinical justification.

Energy efficient convolutional neural networks for arrhythmia detection

Electrocardiograms (ECG) record the heart activity and are the most common and reliable method to detect cardiac arrhythmias, such as atrial fibrillation (AFib). Lately, many commercially available devices such as smartwatches are offering ECG monitoring. Therefore, there is increasing demand for designing deep learning models with the perspective to be physically implemented on these small portable devices with limited energy supply. In this paper, a workflow for the design of small, energy-efficient recurrent convolutional neural network (RCNN) architecture for AFib detection is proposed. However, the approach can be well generalized to every type of long time series. In contrast to previous studies, that demand thousands of additional network neurons and millions of extra model parameters, the logical steps for the generation of a CNN with only 114 trainable parameters are described. The model consists of a small segmented CNN in combination with an optimal energy classifier. The architectural decisions are made by using the energy consumption as a metric in an equally important way as the accuracy. The optimization steps are focused on the software which can be embedded afterwards on a physical chip. Finally, a comparison with some previous relevant studies suggests that the widely used huge CNNs for similar tasks are mostly redundant and unessentially computationally expensive. • We generate energy-efficient 1D-CNN architectures for the detection of heart arrhythmias. • Accuracy and energy are optimized during learning for an accurate light-weight design. • Our model uses 99.4% less trainable parameters in comparison to previous studies. • The method can be applied on every type of continuous time series signals.

Neurologist-Led Management of Implantable Loop-Recorders After Embolic Stroke of Undetermined Source.

IntroductionUpon completion of the workup for stroke, etiology cannot be identified in approximately one-third of the patients, with an embolic stroke of undetermined source (ESUS) accounting for around 50% of these cryptogenic etiologies. Whether management of complex long-term monitoring in order to detect suspected atrial fibrillation (AFib) could be initiated and managed through a neurologist is not sufficiently investigated.Patients and MethodsWe recruited all consecutive patients with ESUS who received implantation after neurological adjudication of Reveal LINQ® loop recorder between January 2016 and July 2020. We collected demographic, clinical, heart- and neuroimaging, laboratory, and electrocardiographic data assessed on prolonged baseline ECG monitoring, number of supraventricular (SVEs) and ventricular (VEs) extrasystolic complexes, and from preimplantation ECG–PQ interval. AFib detection was manually supervised and determined positive when the duration was over 120 s.ResultsWe followed a total of 131 patients for a median of 504 days. There were 45 (34%) manually verified AFib diagnoses. In univariate analysis, earlier implantation after ESUS was associated with AFib detection (13 vs. 31 days, p = 0.011). In multivariate analysis, increased rate of AFib was associated with a more prolonged PQ interval (per 50-ms increase) (HR 1.99, 95% CI 1.39–2.85) and number of SVEs (HR 1.29, 95% CI 1.05–1.57) measured on pre-implantation ECG.ConclusionWe observed similar predictors for Afib after ESUS, albeit with higher frequency than previously reported. This study suggests that the neurologist-led decision, management, and evaluation of ILR after ESUS is feasible.

Abstract 12517: CardioSignal Smartphone Application Detects Atrial Fibrillation in Heart Failure Population

Introduction: The micro-electromechanical sensors of a common smartphone can be used to detect vibrations caused by cardiac motion when the smartphone is placed on chest. With commonly available CardioSignal application (App) the concept has been shown to detect atrial fibrillation (AFib) in clinical setting from a 60 second recording. The CardioSignal smartphone app is a certified CE IIa-class medical device and intended to detect AFib in adult population. AFib and heart failure (HF) frequently coexist and together can lead to increased morbidity. To address this subgroup of individuals with increased risk of Afib, we examined how CardioSignal solution performs in Afib detection in a population of recent HF decompensation. Methods: From a total of 83 HF patients (mean age 64, 70 % men) motion sensor signals were collected and analyzed with the CardioSignal solution. The signals were collected by placing a smartphone on the chest of the patients in supine position. From the recordings two 60 second strips were analyzed. A simultaneous 1-lead ECG data was collected with an ECG patch from the V4 position. The signals were analyzed for quality and rhythm diagnostics from the ECG recording was by a cardiologist blinded for the CardioSignal solution result. Finally a head-to-head comparison was conducted for CardioSignal result and the ECG data. A positive finding for Afib was concluded if both 60 second recordings detected Afib for each person. Results: From the 83 HF patients 166 measurements were recorded. Out of the 166 measurements, 163 passed the quality check for the sensor signals. The preliminary analyses showed sensitivity of the AFib detection algorithm of 80.7% (95% CI 62.5% to 92.6%) and the specificity of 100.0% (93.1% to 100.0%). The positive predictive value was 100.0% and the negative predictive value 89.7% (80.9% to 94.7%). The accuracy was 92.8% (84.9% to 97.3%). Conclusions: A smartphone-based recording and analysis with CardioSignal solution in HF patients appear to be an effective and reliable method to detect possible Afib.