Diagnosing Atrial Fibrillation by Artificial Intelligence

Version	Summary	Created by	Modification	Content Size	Created at	Operation
1		ADRIJA HAJRA	--	1834	2022-04-29 04:03:29	\|
2	format correct	Catherine Yang	Meta information modification	1834	2022-04-29 04:23:02	\|

This entry is adapted from the peer-reviewed paper 10.3390/diagnostics12030689

Atrial fibrillation (AF) is a common arrhythmia affecting 8–10% of the population older than 80 years old. The importance of early diagnosis of atrial fibrillation has been broadly recognized since arrhythmias significantly increase the risk of stroke, heart failure and tachycardia-induced cardiomyopathy with reduced cardiac function. The development of wearable devices has provided a reliable way for healthcare providers to uncover undiagnosed atrial fibrillation in the population, especially those most at risk. Furthermore, with the advancement of artificial intelligence and machine learning, the technology is now able to utilize the database in assisting detection of arrhythmias from the data collected by the devices.

atrial fibrillation artificial intelligence wearable devices machine learning

1. Atrial Fibrillation

Atrial fibrillation (AF) is defined as a type of supraventricular arrhythmia characterized by uncoordinated atrial activation which leads to ineffective atrial contraction ^[1]. It was first described as “auricular fibrillation” by William Harvey in 1628 and was thought to be the dissociation between the peripheral pulse and heartbeat ^[2]. Since then, AF has been studied extensively and has been broadly divided into different categories based on its nature. Paroxysmal AF is defined as AF that terminates spontaneously or with intervention within 7 days of onset. Persistent AF is continuous AF that lasts for more than 7 days ^[3]. Long-standing persistent AF is defined by uninterrupted AF for more than 12 months, while permanent AF is persistent AF with no rhythm control strategy pursued by the patient and physician.

The prevalence of AF increases with aging ^[4]. It is seen in 8–10% of people aged more than 80 years, while it occurs in less than 1% of the population aged 60 to 65 years ^[5]. Females have higher incidental rates compared to males ^[6]. It is also noted that people of European descent are more likely to have AF than African Americans ^[7]. Other risk factors are associated with the development of AF, including hypertension, obesity, diabetes, heart failure, ischemic heart disease, hyperthyroidism, chronic kidney disease, moderate to heavy alcohol use and smoking and sleep disordered breathing ^[8]^[9]^[10]^[11]^[12]^[13]^[14]^[15]^[16].

According to the Centers for Disease Control (CDC) of the United States database, AF contributes to approximately 158,000 deaths each year ^[17]. For more than two decades, the death rate from AF as a primary or contributing cause of death has been rising ^[18]. Each year, AF is the primary diagnosis in more than 454,000 hospitalizations in the US ^[19]. It is estimated that by 2030, about 12.1 million people will be diagnosed with AF in the United States ^[8]. Due to the large burden of this disease with increasing morbidities and mortalities, there have been significant advancements in terms of research for the treatment of AF. Apart from lifestyle modifications for risk reduction, treatment of AF can be categorized into three major categories: rate control, rhythm control and anticoagulation based on the CHA2DS2-VASc score and HAS-BLED score for stroke prevention ^[1]^[15]^[20]. Slowing the ventricular rate using AV nodal blocking agents such as beta blockers, non-dihydropyridine calcium channel blockers (verapamil and diltiazem) and digoxin improves the quality of life and decreases the risk of developing tachycardia-induced cardiomyopathy ^[1]. Alternatively, rhythm-control strategies using pharmacological cardioversion with class 3 anti-arrhythmic drugs based on Vaughan-Williams anti-arrhythmic drug classification, such as amiodarone, electric cardioversion with direct current or catheter ablation of AF foci in the atria, are used to restore and maintain normal sinus rhythm in patients with long-term AF ^[1]^[16]^[21]. In terms of anticoagulation, warfarin is used to prevent thromboembolism in patients with valvular AF, while new oral anticoagulants (NOACs) are often used in patients with non-valvular AF ^[1]. Recent advances in the treatment of AF have shown that left atrial appendage occlusion with the Watchman device is non-inferior to warfarin in preventing stroke in patients with non-valvular AF ^[22]. The medical field is evolving in the management of AF, yet due to the large burden of silent AF, diagnosing AF remains challenging. AF can be asymptomatic or present with symptoms such as palpitations, dyspnea, chest pain, decreased exercise tolerance and fatigue ^[14]. Many patients with silent AF may present with stroke as their first symptom ^[15]. The risk of cerebral embolism increases greatly in chronic AF, which has been estimated to account for approximately 50% of cardioembolic strokes. Atrial fibrillation is also associated with an increased risk of tachycardia-induced cardiomyopathy. Uncontrolled AF can eventually lead to reduced ventricular filling, increased left atrial pressures, hemodynamic instability, reduced cardiac output and morbidity ^[16].

2. Artificial Intelligence and Machine Learning Utilizing Advanced Technology in View of Atrial Fibrillation Diagnosis

2.1. Overview of Algorithms

Artificial intelligence (AI) refers to systems or machines that mimic human intelligence to perform tasks and can iteratively improve themselves based on the information they collect. The term AI is often used interchangeably with its subfields, which include machine learning (ML). Machine learning is focused on building systems that learn or improve based on the data they consume. Categories of ML can be divided into supervised learning, unsupervised learning, semi-supervised learning, reinforcement learning and active learning tasks.

The application of artificial intelligence (AI) and machine learning (ML) in medicine has become important in intense exploration with increased cardiovascular disease, which is responsible for nearly a third of all deaths worldwide ^[23]. Cardiovascular diseases, such as atrial fibrillation (AF), affect up to 34 million people in the world ^[24], and patients with AF exhibit a higher risk of severe health consequences, including death and stroke. There has been considerable research in using machine learning (ML) to improve cardiovascular outcomes in patients. Researchers and clinicians can use AI/ML methods and datasets for diagnosis and disease classification, risk prediction and patient management. Therefore, earlier detection of AF with the use of anticoagulation therapy would mitigate the risk of stroke and other thromboembolic complications.

ML methods involve the scientific study of statistical models and algorithms that can progressively learn from datasets to perform results and achieve goals on a specific task. The processes of the ML pipeline consist of several steps, including data acquisition and preprocessing; feature extraction and selection; and selection of supervised or unsupervised learning methods. ML algorithms can be trained with a small input dataset, and then these trained algorithms can be applied to other large and variable datasets to predict specific cardiovascular diseases. With the introduction of better-trained ML algorithms, more accurate disease prediction would help clinicians obtain an improved diagnosis, classification and risk stratification.

2.2. Supervised Machine Learning for Atrial Fibrillation Detection

Supervised ML is training a model to relate input data and to learn a function that maps input data to target labeled outcomes of interest ^[25]. With a labeled training dataset, the problems can be further categorized into problems of regression and classification. Regression refers to predicting a continuous outcome when target variables are continuous real number values. Classification refers to predicting outcome labels on new data when the target variables are categorical variables. In supervised ML, training data consisting of input features and corresponding data labels (outputs) are provided to an ML algorithm. The ML algorithm then fits the ML model by learning the relationships between the features and the data labels, a process referred to as training. Once the model is trained, it will be able to make predictions from new data, a process referred to as testing. R-R intervals obtained through ECG recording can be translated into a Lorenz plot ^[26]. Subsequently, the data can be further presented to the ML algorithm ^[27].

The feasibility of signal-processed surface electrocardiography (spECG) with the basic use of traditional supervised ML as a diagnostic tool has been used to predict the presence of abnormal cardiac muscle relaxation ^[28]. An ML Cardiio Rhythm algorithm (supervised support vector machine) using facial and fingertip photoplethysmographic (PPG) data obtained from an iPhone camera was tested on 217 cardiology inpatients. The results show that the Cardiio Rhythm algorithm discriminated atrial fibrillation from sinus rhythm with 95% sensitivity, 96% specificity, PPV 92% and NPV 97%. Hence, it is feasible that detection of a facial PPG signal is able to determine pulse irregularity attributable to AF. The result of the study shows high sensitivity and specificity, while a low negative likelihood ratio for AF can be detected from facial PPG signals with the Cardiio Rhythm smartphone application ^[29]. The algorithm of the supervised random forest algorithm was trained using clinical variables from 481 CRT-P patients and tested on 595 CRT-D patients from the COMPANION trial ^[30]. The result shows that death or heart failure hospitalization was predicted within 12 months with an AUC of 0.74. Meanwhile, an eight-fold difference in all-cause mortality was shown between the top and bottom quartiles ^[31]. The application of supervised deep learning with convolutional neural networks (CNNs) was used to train with heart rate, activity level and ECGs from smartwatches of 7500 subjects, with testing on 24 patients. Compared with an insertable cardiac monitor, the smartwatch detected AF with episode sensitivity of 97.5%, PPV of 39.9% and duration sensitivity of 97.7% ^[32]. Analysis incorporates deep neural networks, which are a type of machine learning algorithm with multiple layers of processing that helps to yield higher accuracy in performing pattern recognition of data ^[33]. Supervised deep learning with recurrent neural network (RNN) combined with a supervised gradient boosting classifier was trained using 8183 single-lead ECGs and tested on 3658 subjects. The approach used classified ECG with an overall F1 score of 0.83, which has been validated against the 2017 PhysioNet/CinC Challenge dataset and ranked first in the PhysioNet/CinC competition ^[34]. In total, 30 days of cardiac implantable electronic device remote monitoring data in 3114 non-stroke controls and 71 stroke cases were compared against the past 30 days of remote monitoring before stroke. Three different types of supervised machine learning models were trained, including convolutional neural networks (CNNs), random forest and L1 regularized logistic regression (LASSO), in the study. The result shows that combining CHA2DS2-VASc with CNNs and random forests yielded a validation AUC of 0.696 and test AUC of 0.634, while CHA2DS2-VASc without CNNs and random forests only had an AUC of 0.5 or less in both datasets ^[35].

2.3. Unsupervised Machine Learning for Atrial Fibrillation Detection

Unsupervised ML models learn from clustering and association patterns of unlabeled input data without human intervention. Compared with supervised ML, unsupervised ML is the training of algorithms without definition of the output. Unsupervised ML does not train a model to predict labels from input data. Unsupervised ML instead quantifies natural patterns within input data, and it blinds to the labels of interest. Parsing out these patterns of unsupervised ML reveals an underlying structure to complex data which may help to identify relevant subgroups. One medical example of unsupervised ML was predicting or managing heart disease based on anomalies found within a battery of patient characteristics such as sex, age, body mass index (BMI), lab values and lifestyle, etc., of 9750 patients who participated with smartwatch PPG and tested on 51 patients undergoing cardioversion. The algorithm of the unsupervised approach was used in conjunction with deep learning neural network, followed by supervised classification (semi-supervised). The detection of AF was with an AUC of 0.97 (95% CI, 0.94–1.00; p < 0.001) in patients who underwent cardioversion and an AUC of 0.72 in self-reported ambulatory patients ^[33].

Ebrahimzadeh et al. performed a combination of feature extractions and a mixture of classifier approaches. Their model demonstrated 98.21% accuracy in predicting atrial fibrillation from a large database ^[36]. Machine learning allows large-scale screening and prediction of atrial fibrillation at the primary care setting level ^[37]^[38]^[39].

References

January, C.T.; Wann, L.S.; Calkins, H.; Chen, L.Y.; Cigarroa, J.E.; Cleveland, J.C., Jr.; Ellinor, P.T., Jr.; Ezekowitz, M.D.; Field, M.E.; Furie, K.L.; 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.
Bhatt, H.V.; Fischer, G.W. Atrial Fibrillation: Pathophysiology and Therapeutic Options. J. Cardiothorac. Vasc. Anesth. 2015, 29, 1333–1340.
Hindricks, G.; Potpara, T.; Dagres, N.; Arbelo, E.; Bax, J.J.; Blomström-Lundqvist, C.; Boriani, G.; Castella, M.; Dan, G.-A.; Dilaveris, P.E.; 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. Eur. Heart J. 2021, 42, 4194.
Kornej, J.; Börschel, C.S.; Benjamin, E.J.; Schnabel, R.B. Epidemiology of Atrial Fibrillation in the 21st Century: Novel Methods and New Insights. Circ. Res. 2020, 127, 4–20.
Zimetbaum, P. Atrial Fibrillation. Ann. Intern. Med. 2017, 166, ITC33–ITC48.
Westerman, S.; Wenger, N. Gender Differences in Atrial Fibrillation: A Review of Epidemiology, Management, and Outcomes. Curr. Cardiol. Rev. 2019, 15, 136–144.
Marcus, G.M.; Alonso, A.; Peralta, C.A.; Lettre, G.; Vittinghoff, E.; Lubitz, S.A.; Fox, E.R.; Levitzky, Y.S.; Mehra, R.; Kerr, K.; et al. European Ancestry as a Risk Factor for Atrial Fibrillation in African Americans. Circulation 2010, 122, 2009–2015.
Centers for Disease Control and Prevention. Multiple Cause of Death 1999–2017 on CDC WONDER Online Database, Released December 2018: Data Are from the Multiple Cause of Death Files, 1999–2017, as Compiled from Data Provided by the 57 Vital Statistics Jurisdictions through the Vital Statistics Cooperative Program; Centers for Disease Control and Prevention: Atlanta, GA, USA, 2019.
Benjamin, E.J.; Levy, D.; Vaziri, S.M.; D’Agostino, R.B.; Belanger, A.J.; Wolf, P.A. Independent risk factors for atrial fibrillation in a population-based cohort: The Framingham Heart Study. JAMA 1994, 271, 840–844.
Wang, T.J.; Parise, H.; Levy, D.; D’Agostino, R.B., Sr.; Wolf, P.A.; Vasan, R.S.; Benjamin, E.J. Obesity and the Risk of New-Onset Atrial Fibrillation. JAMA 2004, 292, 2471–2477.
Frost, L.; Vestergaard, P.; Mosekilde, L. Hyperthyroidism and risk of atrial fibrillation or flutter: A population-based study. Arch. Intern. Med. 2004, 164, 1675–1678.
Zhang, D.; Feng, Y.; Leung, F.C.-Y.; Wang, L.; Zhang, Z. Does Chronic Kidney Disease Result in High Risk of Atrial Fibrillation? Front. Cardiovasc. Med. 2019, 6, 82.
Lu, Y.; Guo, Y.; Lin, H.; Wang, Z.; Zheng, L. Genetically determined tobacco and alcohol use and risk of atrial fibrillation. BMC Med. Genom. 2021, 14, 73.
Reynolds, M.R.; Lavelle, T.; Essebag, V.; Cohen, D.J.; Zimetbaum, P. Influence of age, sex, and atrial fibrillation recurrence on quality of life outcomes in a population of patients with new-onset atrial fibrillation: The Fibrillation Registry Assessing Costs, Therapies, Adverse events and Lifestyle (FRACTAL) study. Am. Heart J. 2006, 152, 1097–1103.
Kirchhof, P. The future of atrial fibrillation management: Integrated care and stratified therapy. Lancet 2017, 390, 1873–1887.
Parameswaran, R.; Al-Kaisey, A.M.; Kalman, J.M. Catheter ablation for atrial fibrillation: Current indications and evolving technologies. Nat. Rev. Cardiol. 2021, 18, 210–225.
Centers for Disease Control and Prevention, National Center for Health Statistics. About Multiple Cause of Death, 1999–2019. CDC WONDER Online Database Website; Centers for Disease Control and Prevention: Atlanta, GA, USA, 2019.
Heart Rhythm Society. Complications from Atrial Fibrillation; Heart Rhythm Society: Washington, DC, USA, 2019.
Benjamin, E.J.; Muntner, P.; Alonso, A.; Bittencourt, M.S.; Callaway, C.W.; Carson, A.P.; Chamberlain, A.M.; Chang, A.R.; Cheng, S.; Das, S.R.; et al. Heart disease and stroke statistics—2019 update: A report from the American heart association. Circulation 2019, 139, e56–e528.
Lane, D.A.; Lip, G.Y.H. Use of the CHA 2 DS 2-VASc and HAS-BLED Scores to Aid Decision Making for Thromboprophylaxis in Nonvalvular Atrial Fibrillation. Circulation 2012, 126, 860–865.
Marrouche, N.F.; Brachmann, J.; Andresen, D.; Siebels, J.; Boersma, L.; Jordaens, L.; Merkely, B.; Pokushalov, E.; Sanders, P.; Proff, J.; et al. Catheter Ablation for Atrial Fibrillation with Heart Failure. N. Engl. J. Med. 2018, 378, 417–427.
Reddy, V.Y.; Doshi, S.K.; Kar, S.; Gibson, D.N.; Price, M.J.; Huber, K.; Horton, R.P.; Buchbinder, M.; Neuzil, P.; Gordon, N.T. 5-Year Outcomes After Left Atrial Appendage Closure: From the PREVAIL and PROTECT AF Trials. J. Am. Coll. Cardiol. 2017, 70, 2964–2975.
Deaton, C.; Froelicher, E.S.; Wu, L.H.; Ho, C.; Shishani, K.; Jaarsma, T. The global burden of cardiovascular disease. Eur. J. Cardiovasc. Nurs. 2011, 10 (Suppl. S2), S5–S13.
Chugh, S.S.; Havmoeller, R.; Narayanan, K.; Singh, D.; Rienstra, M.; Benjamin, E.J.; Gillum, R.F.; Kim, Y.H.; McAnulty, J.H., Jr.; Zheng, Z.J.; et al. Worldwide epidemiology of atrial fibrillation: A Global Burden of Disease 2010 Study. Circulation 2014, 129, 837–847.
Sarker, I.H. Machine Learning: Algorithms, Real-World Applications and Research Directions. SN Comput. Sci. 2021, 2, 160.
Esperer, H.D.; Esperer, C.; Cohen, R.J. Cardiac Arrhythmias Imprint Specific Signatures on Lorenz Plots. Ann. Noninvasive Electrocardiol. 2008, 13, 44–60.
Lown, M.; Brown, M.; Brown, C.; Yue, A.M.; Shah, B.N.; Corbett, S.J.; Lewith, G.; Stuart, B.; Moore, M.; Little, P. Machine learning detection of Atrial Fibrillation using wearable technology. PLoS ONE 2020, 15, e0227401.
Sengupta, P.P.; Kulkarni, H.; Narula, J. Prediction of Abnormal Myocardial Relaxation from Signal Processed Surface ECG. J. Am. Coll. Cardiol. 2018, 71, 1650–1660.
Yan, B.P.; Lai, W.H.S.; Chan, C.K.Y.; Chan, S.C.; Chan, L.; Lam, K.; Lau, H.; Ng, C.; Tai, L.; Yip, K.; et al. Contact-Free Screening of Atrial Fibrillation by a Smartphone Using Facial Pulsatile Photoplethysmographic Signals. J. Am. Heart Assoc. 2018, 7.
Bristow, M.R.; Saxon, L.A.; Boehmer, J.; Krueger, S.; Kass, D.A.; De Marco, T.; Carson, P.; Dicarlo, L.; DeMets, D.; White, B.G.; et al. Cardiac-Resynchronization Therapy with or without an Implantable Defibrillator in Advanced Chronic Heart Failure. N. Engl. J. Med. 2004, 350, 2140–2150.
Kalscheur, M.M.; Kipp, R.T.; Tattersall, M.C.; Mei, C.; Buhr, K.A.; DeMets, D.L.; Field, M.E.; Eckhardt, L.L.; Page, C.D. Machine Learning Algorithm Predicts Cardiac Resynchronization Therapy Outcomes: Lessons from the COMPANION Trial. Circ. Arrhythm. Electrophysiol. 2018, 11, e005499.
Wasserlauf, J.; You, C.; Patel, R.; Valys, A.; Albert, D.; Passman, R. Smartwatch Performance for the Detection and Quantification of Atrial Fibrillation. Circ. Arrhythm. Electrophysiol. 2019, 12, e006834.
Tison, G.; Sanchez, J.M.; Ballinger, B.; Singh, A.; Olgin, J.E.; Pletcher, M.J.; Vittinghoff, E.; Lee, E.S.; Fan, S.M.; Gladstone, R.A.; et al. Passive Detection of Atrial Fibrillation Using a Commercially Available Smartwatch. JAMA Cardiol. 2018, 3, 409–416.
Teijeiro, T.; García, C.A.; Castro, D.; Felix, P. Abductive reasoning as a basis to reproduce expert criteria in ECG atrial fibrillation identification. Physiol. Meas. 2018, 39, 084006.
Han, L.; Askari, M.; Altman, R.B.; Schmitt, S.K.; Fan, J.; Bentley, J.P.; Narayan, S.M.; Turakhia, M.P. Atrial Fibrillation Burden Signature and Near-Term Prediction of Stroke: A Machine Learning Analysis. Circ. Cardiovasc. Qual. Outcomes 2019, 12, e005595.
Ebrahimzadeh, E.; Kalantari, M.; Joulani, M.; Shahraki, R.S.; Fayaz, F.; Ahmadi, F. Prediction of paroxysmal Atrial Fibrillation: A machine learning based approach using combined feature vector and mixture of expert classification on HRV signal. Comput. Methods Programs Biomed. 2018, 165, 53–67.
Sekelj, S.; Sandler, B.; Johnston, E.; Pollock, K.G.; Hill, N.R.; Gordon, J.; Tsang, C.; Khan, S.; Ng, F.S.; Farooqui, U. Detecting undiagnosed atrial fibrillation in UK primary care: Validation of a machine learning prediction algorithm in a retrospective cohort study. Eur. J. Prev. Cardiol. 2021, 28, 598–605.
Hill, N.R.; Ayoubkhani, D.; McEwan, P.; Sugrue, D.M.; Farooqui, U.; Lister, S.; Lumley, M.; Bakhai, A.; Cohen, A.T.; O’Neill, M.; et al. Predicting atrial fibrillation in primary care using machine learning. PLoS ONE 2019, 14, e0224582.
Tiwari, P.; Colborn, K.L.; Smith, D.E.; Xing, F.; Ghosh, D.; Rosenberg, M.A. Assessment of a Machine Learning Model Applied to Harmonized Electronic Health Record Data for the Prediction of Incident Atrial Fibrillation. JAMA Netw. Open 2020, 3, e1919396.

© Text is available under the terms and conditions of the Creative Commons Attribution (CC BY) license; additional terms may apply. By using this site, you agree to the Terms and Conditions and Privacy Policy.

Upload a video for this entry

Information

Subjects: Cardiac & Cardiovascular Systems

Contributors MDPI registered users' name will be linked to their SciProfiles pages. To register with us, please refer to https://encyclopedia.pub/register :

View Times: 373

Update Date: 29 Apr 2022

Table of Contents

Video Upload Options

Confirm