Artificial Intelligence-Enabled ECG to Identify Silent Atrial Fibrillation in Embolic Stroke of Unknown Source

Artificial intelligence predicts undiagnosed atrial fibrillation in patients with embolic stroke of undetermined source using sinus rhythm electrocardiograms

BACKGROUND

Artificial intelligence (AI)-enabled sinus rhythm (SR) electrocardiogram (ECG) interpretation can aid in identifying undiagnosed paroxysmal atrial fibrillation (AF) in patients with embolic stroke of undetermined source (ESUS).

OBJECTIVE

The purpose of this study was to assess the efficacy of an AI model in identifying AF based on SR ECGs in patients with ESUS.

METHODS

A transformer-based vision AI model was developed using 737,815 SR ECGs from patients with and without AF to detect current paroxysmal AF or predict the future development of AF within a 2-year period. Probability of AF was calculated from baseline SR ECGs using this algorithm. Its diagnostic performance was further tested in a cohort of 352 ESUS patients from 4 tertiary hospitals, all of whom were monitored using an insertable cardiac monitor (ICM) for AF surveillance.

RESULTS

Over 25.1-month follow-up, AF episodes lasting ≥1 hour were identified in 58 patients (14.4%) using ICMs. In the receiver operating curve (ROC) analysis, the area under the curve for the AI algorithm to identify AF ≥1 hour was 0.806, which improved to 0.880 after integrating the clinical parameters into the model. The AI algorithm exhibited greater accuracy in identifying longer AF episodes (ROC for AF ≥12 hours: 0.837, for AF ≥24 hours: 0.879) and a temporal trend indicating that the AI-based AF risk score increased as the ECG recording approached the AF onset (P for trend <.0001).

CONCLUSIONS

Our AI model demonstrated excellent diagnostic performance in predicting AF in patients with ESUS, potentially enhancing patient prognosis through timely intervention and secondary prevention of ischemic stroke in ESUS cohorts.

Use of Artificial Intelligence in Improving Outcomes in Heart Disease: A Scientific Statement From the American Heart Association

A major focus of academia, industry, and global governmental agencies is to develop and apply artificial intelligence and other advanced analytical tools to transform health care delivery. The American Heart Association supports the creation of tools and services that would further the science and practice of precision medicine by enabling more precise approaches to cardiovascular and stroke research, prevention, and care of individuals and populations. Nevertheless, several challenges exist, and few artificial intelligence tools have been shown to improve cardiovascular and stroke care sufficiently to be widely adopted. This scientific statement outlines the current state of the art on the use of artificial intelligence algorithms and data science in the diagnosis, classification, and treatment of cardiovascular disease. It also sets out to advance this mission, focusing on how digital tools and, in particular, artificial intelligence may provide clinical and mechanistic insights, address bias in clinical studies, and facilitate education and implementation science to improve cardiovascular and stroke outcomes. Last, a key objective of this scientific statement is to further the field by identifying best practices, gaps, and challenges for interested stakeholders.

Artificial intelligence-enabled electrocardiogram (AI-ECG) does not predict atrial fibrillation following patent foramen ovale closure

Background

Atrial fibrillation (AF) is a known complication following patent foramen ovale (PFO) closure. AI-enabled ECG (AI-ECG) acquired during normal sinus rhythm has been shown to identify individuals with AF by noting high-risk ECG features invisible to the human eye. We sought to characterize the value of AI-ECG in predicting AF development following PFO closure and investigate key clinical and procedural characteristics possibly associated with post-procedural AF.

Methods

We performed a retrospective analysis of patients who underwent PFO closure at our hospital from January 2011 to December 2022. We recorded the probability (%) of AF using the Mayo Clinic AI-ECG dashboard from pre- and post-procedure ECGs. The cut-off point of ≥ 11 %, which was found to optimally balance sensitivity and specificity in the original derivation paper (the Youden index) was used to label an AI-ECG "positive" for AF. Pre-procedural transesophageal echocardiography (TEE) and pre- and post-procedure transcranial doppler (TCD) data was also recorded.

Results

Out of 93 patients, 49 (53 %) were male, mean age was 55 ± 15 years with mean post-procedure follow up of 29 ± 3 months. Indication for PFO closure in 69 (74 %) patients was for secondary prevention of transient ischemic attack (TIA) and/or stroke. Twenty patients (22 %) developed paroxysmal AF post-procedure, with the majority within the first month post-procedure (15 patients, 75 %). Patients who developed AF were not significantly more likely to have a positive post-procedure AI-ECG than those who did not develop AF (30 % AF vs 27 % no AF, p = 0.8).Based on the PFO-Associated Stroke Causal Likelihood (PASCAL) classification, patients who had PFO closure for secondary prevention of TIA and/or stroke in the "possible" group were significantly more likely to develop AF than patients in "probable" and "unlikely" groups (p = 0.034). AF-developing patients were more likely to have post-procedure implantable loop recorder (ILR) (55 % vs 9.6 %, p < 0.001), and longer duration of ILR monitoring (121 vs 92.5 weeks, p = 0.035). There were no significant differences in TCD and TEE characteristics, device type, or device size between those who developed AF vs those who did not.

Conclusions

In this small, retrospective study, AI-ECG did not accurately distinguish patients who developed AF post-PFO closure from those who did not. Although AI-ECG has emerged as a valuable tool for risk prediction of AF, extrapolation of its performance to procedural settings such as PFO closure requires further investigation.

Many Models, Little Adoption-What Accounts for Low Uptake of Machine Learning Models for Atrial Fibrillation Prediction and Detection?

(1) Background: Atrial fibrillation (AF) is a major risk factor for stroke and is often underdiagnosed, despite being present in 13-26% of ischemic stroke patients. Recently, a significant number of machine learning (ML)-based models have been proposed for AF prediction and detection for primary and secondary stroke prevention. However, clinical translation of these technological innovations to close the AF care gap has been scant. Herein, we sought to systematically examine studies, employing ML models to predict incident AF in a population without prior AF or to detect paroxysmal AF in stroke cohorts to identify key reasons for the lack of translation into the clinical workflow. We conclude with a set of recommendations to improve the clinical translatability of ML-based models for AF. (2) Methods: MEDLINE, Embase, Web of Science, Clinicaltrials.gov, and ICTRP databases were searched for relevant articles from the inception of the databases up to September 2022 to identify peer-reviewed articles in English that used ML methods to predict incident AF or detect AF after stroke and reported adequate performance metrics. The search yielded 2815 articles, of which 16 studies using ML models to predict incident AF and three studies focusing on ML models to detect AF post-stroke were included. (3) Conclusions: This study highlights that (1) many models utilized only a limited subset of variables available from patients' health records; (2) only 37% of models were externally validated, and stratified analysis was often lacking; (3) 0% of models and 53% of datasets were explicitly made available, limiting reproducibility and transparency; and (4) data pre-processing did not include bias mitigation and sufficient details, leading to potential selection bias. Low generalizability, high false alarm rate, and lack of interpretability were identified as additional factors to be addressed before ML models can be widely deployed in the clinical care setting. Given these limitations, our recommendations to improve the uptake of ML models for better AF outcomes include improving generalizability, reducing potential systemic biases, and investing in external validation studies whilst developing a transparent modeling pipeline to ensure reproducibility.

Atrial fibrillation and stroke: State-of-the-art and future directions

Atrial fibrillation (AF) and stroke remain a major cause of morbidity and mortality. The two conditions shared common co-morbidities and risk factors. AF-related strokes are associated with worse clinical outcomes and higher mortality compared to non-AF-related. Early detection of AF is vital for prevention. While various scores have been developed to predict AF in such a high-risk group, they are yet to incorporated into clinical guidelines. Novel markers and predictors of AF including coronary and intracranial arterial calcification have also been studied. There are also ongoing debates on the management of acute stroke in patients with AF, and those who experienced breakthrough stroke while on oral anticoagulants. We provided an overview of the complex interplay between AF and stroke, as well as the treatment and secondary prevention of stroke in AF. We also comprehensively discussed the current evidence and the ongoing conundrums, and highlighted the future directions on the topic.

Advances in Cardiac Workup for Transient Ischemic Attack: Improving Diagnostic Yield and Reducing Recurrent Stroke Risk

Transient ischemic attack (TIA) is a warning sign for an impending stroke, with a 10-20% chance of a stroke occurring within 90 days of the initial event. Current clinical practice for cardiac workup in TIA includes cardiac enzymes, with 12-lead electrocardiogram, transthoracic echocardiography, and 24-hour Holter monitoring. However, the diagnostic yield of these investigations is variable, and there is a need for better diagnostic approaches to increase the detection of cardiac abnormalities in a cost-effective way. This review article examines the latest research on emerging diagnostic tools and strategies and discusses the potential benefits and challenges of using these advanced diagnostic approaches in clinical practice. Novel biomarkers, imaging techniques, and prolonged rhythm monitoring devices have shown great promise in enhancing the diagnostic yield of cardiac workup in TIA patients. Echocardiography, Transcranial Doppler ultrasound, cardiac MRI, and cardiac CT are among the promising diagnostic tools being studied. We conclude the article with a suggested diagnostic algorithm for cardiac workup in TIA. Further research is necessary to enhance their usefulness and to outline future directions for research and clinical practice in this field.

Artificial Intelligence for the Detection and Treatment of Atrial Fibrillation

AF is the most common clinically relevant cardiac arrhythmia associated with multiple comorbidities, cardiovascular complications (e.g. stroke) and increased mortality. As artificial intelligence (AI) continues to transform the practice of medicine, this review article highlights specific applications of AI for the screening, diagnosis and treatment of AF. Routinely used digital devices and diagnostic technology have been significantly enhanced by these AI algorithms, increasing the potential for large-scale population-based screening and improved diagnostic assessments. These technologies have similarly impacted the treatment pathway of AF, identifying patients who may benefit from specific therapeutic interventions. While the application of AI to the diagnostic and therapeutic pathway of AF has been tremendously successful, the pitfalls and limitations of these algorithms must be thoroughly considered. Overall, the multifaceted applications of AI for AF are a hallmark of this emerging era of medicine.

Machine Learning and the Conundrum of Stroke Risk Prediction

Stroke is a leading cause of death worldwide. With escalating healthcare costs, early non-invasive stroke risk stratification is vital. The current paradigm of stroke risk assessment and mitigation is focused on clinical risk factors and comorbidities. Standard algorithms predict risk using regression-based statistical associations, which, while useful and easy to use, have moderate predictive accuracy. This review summarises recent efforts to deploy machine learning (ML) to predict stroke risk and enrich the understanding of the mechanisms underlying stroke. The surveyed body of literature includes studies comparing ML algorithms with conventional statistical models for predicting cardiovascular disease and, in particular, different stroke subtypes. Another avenue of research explored is ML as a means of enriching multiscale computational modelling, which holds great promise for revealing thrombogenesis mechanisms. Overall, ML offers a new approach to stroke risk stratification that accounts for subtle physiologic variants between patients, potentially leading to more reliable and personalised predictions than standard regression-based statistical associations.

Probability of Cardioembolic vs. Atherothrombotic Pathogenesis of Cryptogenic Strokes in Older Patients

Some clinical, laboratory, ECG, and echocardiographic parameters could provide useful indications to assess the probability of cardioembolism or atherothrombosis in cryptogenic strokes. We retrospectively examined 290 patients with ischemic stroke aged ≥60 years, divided into 3 groups: strokes originating from large artery atherothrombosis (n = 92), cardioembolic strokes caused by paroxysmal atrial fibrillation (n = 88) and cryptogenic strokes (n = 110). In addition to echocardiographic and routine clinical-laboratory variables, neutrophil:lymphocyte ratio, red blood cell distribution width, mean platelet volume, P wave and PR interval duration and biphasic inferior P waves, both on admission and after 7 to 10 days, were also considered. By multiple logistic regression, cardioembolic strokes were compared with large artery atherothrombosis strokes, and beta coefficients were rounded to produce a scoring system. Late PR interval ≥188 ms, left atrium ≥4 cm, left ventricular end-diastolic volume <65 ml, and posterior circulation syndrome were associated with paroxysmal atrial fibrillation (positive scores). In contrast, male gender, hypercholesterolemia, and initial platelet count ≥290 × 10⁹/L were associated with atherothrombosis of large arteries (negative scores). The algebraic sum of these scores produced values indicative of cardioembolism if >0 (positive predictive value 89.1%), or of atherothrombosis, if ≤0 (positive predictive value 72.5%). The area under the receiver operating characteristic curve was 0.85. Among cryptogenic strokes, 41.5% had a score >0 (probable atrial fibrillation) and 58.5% had a score ≤0 (possible atherothrombosis). In conclusion, a scoring system based on electrocardiogram, laboratory, clinical and echocardiographic parameters can provide useful guidance for further investigations and secondary prevention in older patients with cryptogenic stroke.

Electrocardiographic P terminal force in lead V1, its components, and the association with stroke and atrial fibrillation or flutter

BACKGROUND

The electrocardiographic (ECG) marker P terminal force V1 (PTFV1) is generally perceived as a marker of left atrial pathology and has been associated with atrial fibrillation or flutter (AF).

OBJECTIVE

The purpose of this study was to determine the association between PTFV1 components (duration and amplitude) and incident AF and stroke/transient ischemic attack (TIA).

METHODS

The study included patients with an ECG recorded at the Copenhagen General Practitioners Laboratory in 2001 to 2011. PTFV1 ≥4 mV·ms was considered abnormal. Patients with abnormal PTFV1 were stratified into tertiles based on duration (PTDV1) and amplitude (PTAV1) values. Cox regressions adjusted for age, sex, and relevant comorbidities were used to investigate associations between abnormal PTFV1 components and AF and stroke/TIA.

RESULTS

Of 267,636 patients, 5803 had AF and 18,176 had stroke/TIA (follow-up 6.5 years). Abnormal PTFV1 was present in 44,549 subjects (16.7%) and was associated with an increased risk of AF and stroke/TIA. Among patients with abnormal PTFV1, the highest tertile of PTDV1 (78-97 ms) was associated with the highest risk of AF (hazard ratio [HR] 1.37; 95% confidence interval [CI] 1.23-1.52) and highest risk of stroke/TIA (HR 1.13; 95% CI 1.05 -1.20). For PTAV1, the highest tertile (78-126 μV) conferred the highest risk of AF and stroke/TIA (HR 1.20; 95% CI 1.09-1.32; and HR 1.21; 95% CI 1.14-1.25, respectively).

CONCLUSION

Abnormal PTFV1 was associated with an increased risk of AF and stroke/TIA. Increasing PTDV1 showed a dose-response relationship with the development of AF and stroke/TIA, whereas the association between PTAV1 and AF was less apparent.

Continuous ECG monitoring should be the heart of bedside AI-based predictive analytics monitoring for early detection of clinical deterioration

The idea that we can detect subacute potentially catastrophic illness earlier by using statistical models trained on clinical data is now well-established. We review evidence that supports the role of continuous cardiorespiratory monitoring in these predictive analytics monitoring tools. In particular, we review how continuous ECG monitoring reflects the patient and not the clinician, is less likely to be biased, is unaffected by changes in practice patterns, captures signatures of illnesses that are interpretable by clinicians, and is an underappreciated and underutilized source of detailed information for new mathematical methods to reveal.

Clinical significance, challenges and limitations in using artificial intelligence for electrocardiography-based diagnosis

Cardiovascular diseases are one of the leading global causes of mortality. Currently, clinicians rely on their own analyses or automated analyses of the electrocardiogram (ECG) to obtain a diagnosis. However, both approaches can only include a finite number of predictors and are unable to execute complex analyses. Artificial intelligence (AI) has enabled the introduction of machine and deep learning algorithms to compensate for the existing limitations of current ECG analysis methods, with promising results. However, it should be prudent to recognize that these algorithms also associated with their own unique set of challenges and limitations, such as professional liability, systematic bias, surveillance, cybersecurity, as well as technical and logistical challenges. This review aims to increase familiarity with and awareness of AI algorithms used in ECG diagnosis, and to ultimately inform the interested stakeholders on their potential utility in addressing present clinical challenges.

Evaluating the Risk of Paroxysmal Atrial Fibrillation in Noncardioembolic Ischemic Stroke Using Artificial Intelligence-Enabled ECG Algorithm

Background

The identification of latent atrial fibrillation (AF) in patients with ischemic stroke (IS) attributed to noncardioembolic etiology may have therapeutic implications. An artificial intelligence (AI) model identifying the electrocardiographic signature of AF present during normal sinus rhythm (NSR; AI-ECG-AF) can identify individuals with a high likelihood of paroxysmal AF (PAF) with NSR electrocardiogram (ECG).

Objectives

Using AI-ECG-AF, we aimed to compare the PAF risk between noncardioembolic IS subgroups and general patients of a university hospital after controlling for confounders. Further, we sought to compare the risk of PAF among noncardioembolic IS subgroups.

Methods

After training AI-ECG-AF with ECG data of university hospital patients, model inference outputs were obtained for the control group (i.e., general patient population) and NSRs of noncardioembolic IS patients. We conducted multiple linear regression (MLiR) and multiple logistic regression (MLoR) analyses with inference outputs (for MLiR) or their binary form (set at threshold = 0.5 for MLoR) used as dependent variables and patient subgroups and potential confounders (age and sex) set as independent variables.

Results

The number of NSRs inferenced for the control group, cryptogenic, large artery atherosclerosis (LAA), and small artery occlusion (SAO) strokes were 133,340, 133, 276, and 290, respectively. The regression analyses indicated that patients with noncardioembolic IS had a higher PAF risk based on AI-ECG-AF relative to the control group, after controlling for confounders with the “cryptogenic” subgroup having the highest risk (odds ratio [OR] = 1.974, 95% confidence interval [CI]: 1.371–2.863) followed by the “LAA” (OR = 1.592, 95% CI: 1.238–2.056) and “SAO” subgroups (OR = 1.400, 95% CI: 1.101–1.782). Subsequent regression analyses failed to illustrate the differences in PAF risk based on AI-ECG-AF among noncardioembolic IS subgroups.

Conclusion

Using AI-ECG-AF, we found that noncardioembolic IS patients had a higher PAF risk relative to the general patient population. The results from our study imply the need for more vigorous cardiac monitoring in noncardioembolic IS patients. AI-ECG-AF can be a cost-effective screening tool to identify high-risk noncardioembolic IS patients of PAF on-the-spot to be candidates for receiving additional prolonged cardiac monitoring. Our study highlights the potential of AI in clinical practice.

Artificial Intelligence and Atrial Fibrillation

In the context of atrial fibrillation (AF), traditional clinical practices have thus far fallen short in several domains such as identifying patients at risk of incident AF or patients with concomitant undetected paroxysmal AF. Novel approaches leveraging artificial intelligence have the potential to provide new tools to deal with some of these old problems. In this review we focus on the roles of artificial intelligence-enabled ECG pertaining to AF, potential roles of deep learning (DL) models in the context of current knowledge gaps, as well as limitations of these models. One key area where DL models can translate to better patient outcomes is through automated ECG interpretation. Further, we overview some of the challenges facing AF screening and the harms and benefits of screening. In this context, a unique model was developed to detect underlying hidden AF from sinus rhythm and is discussed in detail with its potential uses. Knowledge gaps also remain regarding the best ways to monitor patients with embolic stroke of undetermined source (ESUS) and who would benefit most from oral anticoagulation. The AI-enabled AF model is one potential way to tackle this complex problem as it could be used to identify a subset of high-risk ESUS patients likely to benefit from empirical oral anticoagulation. Role of DL models assessing AF burden from long duration ECG data is also discussed as a way of guiding management. There is a trend towards the use of consumer-grade wristbands and watches to detect AF from photoplethysmography data. However, ECG currently remains the gold standard to detect arrythmias including AF. Lastly, role of adequate external validation of the models and clinical trials to study true performance is discussed. This article is protected by copyright. All rights reserved.

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.