Related Articles: Diagnostic ECG
Atrial fibrillation is an underlying cause of major morbidity and mortality across the globe. According to a review article in Stroke, more than five million people in the United States were estimated to have Afib in 2010, and the number exceeded eight million in the European Union.1 The prevalence of the arrhythmia is expected to rise over the coming decades due to both aging populations and enhanced atrial fibrillation screening efforts. Projections put the number of people with Afib at more than 12 million in the US by 2030 and 17.9 million in the EU by 2060.1 The anticipated growth in Afib prevalence lends even greater urgency to early ECG detection in preventing complications such as cryptogenic stroke.
Cryptogenic Stroke and Underlying Atrial Fibrillation
As the Stroke review article notes, the presence of Afib quintuples the risk of stroke. A review article in the Journal of Atrial Fibrillation states that Afib is a factor in at least 15% of all strokes.2 Particularly insidious is the occurrence of asymptomatic Afib, which an evaluation in American Family Physician found to affect as many as 59% of participants.3
Asymptomatic or previously undiagnosed Afib can result in significant—and potentially preventable—morbidity, disability, and even death, leading international societies to underscore the importance of taking measures to find it.
The American Heart Association (AHA), American College of Cardiology (ACC), and Heart Rhythm Society's (HRS) Guideline for the Management of Patients With Atrial Fibrillation states that "clinically unrecognized and asymptomatic AF is a potentially important cause of stroke, supporting efforts for early detection of AF in at-risk individuals." 4 In a 2019 update to that guidance, a recommendation indicates that for patients with cryptogenic stroke who have inconclusive external ambulatory monitoring, implantation of a cardiac monitor (a loop recorder to look for silent Afib) is reasonable.5
Cryptogenic stroke refers to cerebral infarction due to an obscure or idiopathic mechanism. According to a Frontiers in Physiology review article, as many as 25% to 40% of ischemic strokes are cryptogenic, and the etiology for these cryptogenic strokes is often never identified.6 However, additional extended ECG monitoring may be able to identify paroxysmal Afib in a subset of cryptogenic strokes.
ECG Screening for Afib
The AHA/ACC/HRS guidelines state that "Prolonged or frequent monitoring may be necessary to reveal episodes of asymptomatic Afib," and the ACC recommends considering prolonged cardiac monitoring beyond 24-hour telemetry as part of the cardiac assessment for cryptogenic stroke.7
The European Afib guidelines also weigh in on atrial fibrillation screening. The authors recommend looking for the arrhythmia with short-term ECG recordings followed by continuous ECG monitoring for at least 72 hours, "also considering a tiered longer ECG monitoring approach and insertion of an intracardiac monitor in case of cryptogenic stroke." 8
A Cochrane Review concluded that both systematic and opportunistic screening for Afib increase detection rates of asymptomatic Afib, based on existing evidence.9 Ambulatory ECG monitoring is the most common modality for asymptomatic Afib detection and screening, particularly following stroke or transient ischemic attacks, and various methods are currently available. These include Holter monitors, event recorders (external as well as implantable), ECG patch recorders, and most recently, three-lead and single-lead ECG monitoring.
A feasibility study in the International Journal of Cardiology evaluated the use of three-lead ECG monitoring as a screening tool for Afib and found that the three-lead technology was feasible and effective.10 Another study in Europace looked at the yield of Afib screening with a handheld, single-lead ECG monitor and found use of the single-lead device to be feasible and high-yield, resulting in a single-day yield of 1.1% new cases of Afib among a primary care population presenting for routine influenza vaccination.11
The Promise of AI
Ever more promising opportunities for reducing risk of cryptogenic stroke due to asymptomatic Afib are being realized through advances in artificial intelligence (AI) for cardiac care. Indeed, the European Afib guidelines acknowledge the potential for new digital ECG analysis methods like machine learning and AI, along with consumer wearables and other innovative technologies, to aid in the detection of arrhythmia. "These innovations may help to personalize therapy and risk stratification," the guideline authors state, while also noting that "studies are needed to evaluate such opportunities and to define for which groups of patients this is worthwhile." 8
The Need for Large Data Sets
As this technology evolves, large data sets are necessary to obtain enough information for thorough evaluation.
A study in the Lancet took advantage of just such a data set to assess the ability of a newly developed AI-enabled ECG algorithm to identify patients with Afib using information gathered in their ECG during normal sinus rhythm.12 The study, which ran from 1993 to 2017 and included over 180,000 patients, collected nearly 650,000 normal sinus ECGs for analysis. Each ECG was obtained using a standard 10-second, 12-lead GE ECG, with each patient in supine position.
Following data collection, each ECG was initially read by the GE ECG, and the interpretation was verified or corrected by a physician-supervised, trained ECG technician. Patients with at least one ECG showing Afib or atrial flutter were classified as positive for Afib.
After training and internal validation of the AI-enabled algorithm, the researchers tested its ability to identify patients who had a prior history of Afib. In the testing cohort, 8.4% of patients had verified Afib before the collection of the normal sinus rhythm ECG, and the AI-enabled algorithm detected this group with 83.3% accuracy.12
To learn more about the power of the ECG in today's clinical landscape, browse our Diagnostic ECG Clinical Insights Center.
AI-Enabled ECG Software can Improve Rhythm Diagnoses
The aforementioned study, however, represents only one type of algorithm on the market today. For instance, at Helsinki University Hospital in Finland, which treats about half a million patients each year, physicians started using CardioDay Holter ECG software with the CARESCAPE monitoring and telemetry network to make the process of atrial fibrillation screening in hospitalized patients more efficient and less burdensome.13
And it worked. Use of the software allowed inpatient ECGs to be examined quickly by a clinical physiologist for Afib, with the process taking about 5 minutes (down from about 1 hour). Overall, it's estimated that the strategy saves the equivalent of about one working day for a medical technician each week, leading the hospital to conclude that it's "convenient and cost-effective."13
AI-enabled ECG software can improve rhythm diagnoses in a multitude of ways, offering automatic detection of Afib episodes to help prevent cryptogenic strokes. Additionally, the enhanced ability to detect occult Afib can help reduce the development of tachycardia-induced cardiomyopathy, which, as the AHA/ACC/HRS guidelines observe, can otherwise occur in untreated patients who have Afib with rapid ventricular response.
A recent review in Circulation Research suggests that larger health systems may use AI-enabled ECG algorithms as a low-cost, mass screening tool for Afib and other cardiovascular issues.14 And, as the authors of a review in Mayo Clinic Proceedings note, ECG technology is becoming cheaper and more accessible to the general population.15 AI could eventually be used to interpret ECG data collected from personal devices, predicting rhythm disorders like Afib and triaging patients to appropriate medical personnel.
While AI-enabled algorithms may be used to enhance current diagnostic methods, it is unlikely that they will replace standard practices in the near term. As the technology develops, external automatic monitoring of ECG through the use of AI will enhance cardiologists' ability to detect and treat asymptomatic Afib, improving patient outcomes related to cryptogenic stroke and underlying atrial fibrillation. Ultimately, AI will increase efficiency and effectiveness in clinical practice and enable the delivery of a more advanced level of personalized medicine.
References:
1. Passman R, Bernstein RA. New appraisal of atrial fibrillation burden and stroke prevention. Stroke. 2016;47:570-576. https://www.ahajournals.org/doi/10.1161/STROKEAHA.115.009930
2. Majos E, Dabrowski R. Significance and management strategies for patients with asymptomatic atrial fibrillation. Journal of Atrial Fibrillation. 2015;7(5):1169. http://www.jafib.com/published.php?type=full&id=1169
3. Newell A, Hayes J, Smith R. Evaluation of asymptomatic atrial fibrillation. American Family Physician. 2012;86(6):Online. https://www.aafp.org/pubs/afp/issues/2012/0915/od1.html
4. January CT, Wann LS, Alpert JS, et al. 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines and the Heart Rhythm Society. Journal of the American College of Cardiology. 2014;64(21):e1-e76. https://www.jacc.org/doi/10.1016/j.jacc.2014.03.022
5. January CT, Wann LS, Calkins H, et al. 2019 AHA/ACC/HRS focused update of the 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines and the Heart Rhythm Society in collaboration with the Society of Thoracic Surgeons. Circulation. 2019;140:e125-e151. https://www.ahajournals.org/doi/10.1161/CIR.0000000000000665
6. Andrade JG, Field T, Khairy P. Detection of occult atrial fibrillation in patients with embolic stroke of uncertain source: a work in progress. Frontiers in Physiology. 2015;6:100. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4381503/
7. Serhal M. Evaluation of cryptogenic stroke. ACC.org. https://www.acc.org/latest-in-cardiology/articles/2019/10/10/23/20/evaluation-of-cryptogenic-stroke. Accessed August 22, 2022.
8. Hindricks G, Potpara T, Dagres N, et al. 2020 ESC guidelines for the diagnosis and management of atrial fibrillation developed in collaboration with the European Association for Cardio-Thoracic Surgery (EACTS): the task force for the diagnosis and management of atrial fibrillation of the European Society of Cardiology (ESC) developed with the special contribution of the European Heart Rhythm Association (EHRA) of the ESC. European Heart Journal. 2020;42(5):373-498. https://academic.oup.com/eurheartj/article/42/5/373/5899003
9. Moran PS, Teljeur C, Ryan M, Smith SM. Systematic screening for the detection of atrial fibrillation. Cochrane Database of Systematic Reviews. 2016;2016(6):CD009586. https://pubmed.ncbi.nlm.nih.gov/27258214/
10. Bury G, Swan D, Cullen W, et al. Screening for atrial fibrillation in general practice: A national, cross-sectional study of an innovative technology. International Journal of Cardiology. 2015;178:247-252. https://www.internationaljournalofcardiology.com/article/S0167-5273(14)01974-3/fulltext
11. Kaasenbrood F, Hollander M, Rutten FH, et al. Yield of screening for atrial fibrillation in primary care with a hand-held, single-lead electrocardiogram device during influenza vaccination. Europace. 2016;18(10):1514-1520. https://academic.oup.com/europace/article/18/10/1514/2468565
12. Attia ZI, Noseworthy PA, Lopez-Jimenez F, et al. An artificial intelligence-enabled ECG algorithm for the identification of patients with atrial fibrillation during sinus rhythm: a retrospective analysis of outcome prediction. Lancet. 2019;394(10201):861-867. https://www.thelancet.com/journals/lancet/article/PIIS0140-6736%2819%2931721-0/fulltext
13. GE. "Convenient and cost effective": atrial fibrillation screening with CardioDay and CARESCAPE monitors. GEHealthcare.com.au. https://www.gehealthcare.com.au/-/jssmedia/documents/uk-doc/dcar--eu--cardioday--carescape-afib--testimonial--english--052019--jb67455xe.pdf?rev=-1. Accessed August 23, 2022.
14. Siontis KC, Yao X, Pirruccello JP, et al. How will machine learning inform the clinical care of atrial fibrillation? Circulation Research. 2020;127(1):155-169. https://www.ahajournals.org/doi/10.1161/CIRCRESAHA.120.316401
15. Lopez-Jimenez F, Attia Z, Arruda-Olson AM, et al. Artificial intelligence in cardiology: present and future. Mayo Clinic Proceedings. 2020;95(5):1015-1039. https://www.mayoclinicproceedings.org/article/S0025-6196%2820%2930138-5/fulltext
