Abstract
Background
As a major complication of non-valvular atrial fibrillation (NVAF), left atrial appendage
(LAA) thrombosis is associated with cerebral ischemic strokes, as well as high morbidity.
Due to insufficient incorporation of risk factors, most current scoring methods are
limited to the analysis of relationships between clinical characteristics and LAA
thrombosis rather than detecting potential risk. Therefore, this study proposes a
clinical data-driven machine learning method to predict LAA thrombosis of NVAF.
Methods
Patients with NVAF from January 2014 to June 2022 were enrolled from 40 Southwest
Hospital. We selected 40 variables for analysis, including demographic data, medical
history records, laboratory results, and the structure of LAA. Three machine learning
algorithms were adopted to construct classifiers for the prediction of LAA thrombosis
risk. The most important variables related to LAA thrombosis and their influences
were recognized by SHapley Addictive exPlanations method. In addition, we compared
our model with CHADS2 and CHADS2-VASc scoring methods.
Results
A total of 713 participants were recruited, including 127 patients with LAA thrombosis
and 586 patients with no obvious thrombosis. The consensus models based on Random
Forest and eXtreme Gradient Boosting LAA thrombosis prediction (RXTP) achieved the
best accuracy of 0.865, significantly outperforming CHADS2 score and CHA2DS2-VASc
score (0.754 and 0.754, respectively). The SHAP results showed that B-type natriuretic
peptide, left atrial appendage width, C-reactive protein, Fibrinogen and estimated
glomerular filtration rate are closely related to the risk of LAA thrombosis in nonvalvular
atrial fibrillation.
Conclusions
The RXTP-NVAF model is the most effective model with the greatest ROC value and recall
rate. The summarized risk factors obtained from SHAP enable the optimization of the
treatment strategy, thereby preventing thromboembolism events and the occurrence of
cardiogenic ischemic stroke.
Keywords
To read this article in full you will need to make a payment
Purchase one-time access:
Academic & Personal: 24 hour online accessCorporate R&D Professionals: 24 hour online accessOne-time access price info
- For academic or personal research use, select 'Academic and Personal'
- For corporate R&D use, select 'Corporate R&D Professionals'
Subscribe:
Subscribe to Thrombosis ResearchAlready a print subscriber? Claim online access
Already an online subscriber? Sign in
Register: Create an account
Institutional Access: Sign in to ScienceDirect
References
- Atrial fibrillation.Nat. Rev. Dis. Primers. 2022; 8: 21https://doi.org/10.1038/s41572-022-00347-9
- Atrial fibrillation prevalence, awareness and management in a nationwide survey of adults in China.Heart. 2021; 107: 535-541https://doi.org/10.1136/heartjnl-2020-317915
- Global epidemiology of atrial fibrillation: an increasing epidemic and public health challenge.Int. J. Stroke. 2021; 16: 217-221https://doi.org/10.1177/1747493019897870
- Atrial fibrillation.Ann. Intern. Med. 2017; 166: ITC33-ITC48https://doi.org/10.7326/AITC201703070
- Contemporary atrial fibrillation management: a comparison of the current AHA/ACC/HRS, CCS, and ESC guidelines.Can J Cardiol. 2017; 33: 965-976https://doi.org/10.1016/j.cjca.2017.06.002
- Stroke prevention in atrial fibrillation–an asian stroke perspective.Heart Rhythm. 2013; 10: 1082-1088https://doi.org/10.1016/j.hrthm.2013.03.017
- Using artificial intelligence with natural language processing to combine electronic health Record's structured and free text data to identify nonvalvular atrial fibrillation to decrease strokes and death: evaluation and case-control study.J. Med. Internet Res. 2021; 23e28946https://doi.org/10.2196/28946
- Structure and function of the left atrium and left atrial appendage: AF and stroke implications.J. Am. Coll. Cardiol. 2017; 70: 3157-3172https://doi.org/10.1016/j.jacc.2017.10.063
- Numerical analysis of hemodynamic changes in the left atrium due to atrial fibrillation.J. Biomech. 2015; 48: 472-478https://doi.org/10.1016/j.jbiomech.2014.12.025
- Mechanisms of thrombogenesis in atrial fibrillation: Virchow's triad revisited.Lancet (London, England). 2009; 373: 155-166https://doi.org/10.1016/s0140-6736(09)60040-4
- Endothelial dysfunction and pathophysiological correlates in atrial fibrillation.Heart. 2009; 95: 102-106https://doi.org/10.1136/hrt.2007.135277
- Transesophageal echocardiographic correlates of thromboembolism in high-risk patients with nonvalvular atrial fibrillation. The Stroke Prevention in Atrial Fibrillation Investigators Committee on Echocardiography.Ann. Intern. Medicine. 1998; 128: 639-647https://doi.org/10.7326/0003-4819-128-8-199804150-00005
- Importance of left atrial appendage flow as a predictor of thromboembolic events in patients with atrial fibrillation.Eur. Heart J. 1999; 20: 979-985https://doi.org/10.1053/euhj.1998.1453
- Validation of clinical classification schemes for predicting stroke: results from the National Registry of atrial fibrillation.JAMA. 2001; 285: 2864-2870https://doi.org/10.1001/jama.285.22.2864
- Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation.Chest. 2010; 137: 263-272https://doi.org/10.1378/chest.09-1584
- Analysis of risk factors for thrombosis of the left atrium/left atrial appendage in patients with non-valvular atrial fibrillation.Cardiovasc. J. Afr. 2021; 32: 116-122https://doi.org/10.5830/CVJA-2019-071
- Impact of left atrial appendage location on risk of thrombus formation in patients with atrial fibrillation.Biomech. Model. Mechanobiol. 2021; 20: 1431-1443https://doi.org/10.1007/s10237-021-01454-4
- The refinement of risk stratification for atrial thrombus or spontaneous Echo contrast in nonvalvular atrial fibrillation.Int. Heart J. 2017; 58: 885-893https://doi.org/10.1536/ihj.16-444
- Personalized anti-thrombotic management of patients with non-valvular atrial fibrillation and a CHA2DS2-VASc score of 1 — a statement of the ESC working group on cardiovascular pharmacotherapy and ESC Council on stroke [corrected].Eur. Heart J. 2021; 42: 541-543https://doi.org/10.1093/eurheartj/ehaa1081
- CHADS-VASc and has-BLED scores do not accurately stratify risk for stroke or bleed in fall victims with atrial fibrillation.Am. J. Emerg. Med. 2022; 51: 119-123https://doi.org/10.1016/j.ajem.2021.10.008
- Machine learning in medicine.Circulation. 2015; 132: 1920-1930https://doi.org/10.1161/CIRCULATIONAHA.115.001593
- Artificial intelligence in healthcare: past, present and future.Stroke Vasc. Neurol. 2017; 2: 230-243https://doi.org/10.1136/svn-2017-000101
- Machine learning models for predicting in-hospital mortality in acute aortic dissection patients.Front. Cardiovasc. Med. 2021; 8727773https://doi.org/10.3389/fcvm.2021.727773
- Prediction of the development of acute kidney injury following cardiac surgery by machine learning.Crit. Care. 2020; 24: 478https://doi.org/10.1186/s13054-020-03179-9
- Machine learning and prediction of all-cause mortality in COPD.Chest. 2020; 158: 952-964https://doi.org/10.1016/j.chest.2020.02.079
- KNIME for reproducible cross-domain analysis of life science data.J. Biotechnol. 2017; 261: 149-156https://doi.org/10.1016/j.jbiotec.2017.07.028
- How can SHAP values help to shape metabolic stability of chemical compounds?.J Cheminform. 2021; 13: 74https://doi.org/10.1186/s13321-021-00542-y
- 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: executive summary: a report of the American College of Cardiology/American Heart Association Task Force on practice guidelines and the Heart Rhythm Society.Circulation. 2014; 130: 2071-2104https://doi.org/10.1161/CIR.0000000000000040
- Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation.Chest. 2010; 137: 263-272https://doi.org/10.1378/chest.09-1584
- External validation of a novel transthoracic echocardiographic tool in predicting left atrial appendage thrombus formation in patients with nonvalvular atrial fibrillation.Eur. Heart J. Cardiovasc. Imaging. 2013; 14: 876-881https://doi.org/10.1093/ehjci/jes313
- Impact of diastolic function parameters on the risk for left atrial appendage thrombus in patients with nonvalvular atrial fibrillation: a prospective study.J. Am. Soc. Echocardiogr. 2016; 29: 545-553https://doi.org/10.1016/j.echo.2016.01.014
- The left atrial appendage: anatomy, function, and noninvasive evaluation.JACC Cardiovasc. Imaging. 2014; 7: 1251-1265https://doi.org/10.1016/j.jcmg.2014.08.009
- The left atrial appendage: from embryology to prevention of thromboembolism.Eur. Heart J. 2017; 38: 877-887https://doi.org/10.1093/eurheartj/ehw159
- Evaluating the novel parameters for assessing the LAA function and thrombus formation with nonvalvular atrial fibrillation.SaudiJ. Biol. Sci. 2021; 28: 560-565https://doi.org/10.1016/j.sjbs.2020.10.041
- The value of diastolic function parameters in the prediction of left atrial appendage thrombus in patients with nonvalvular atrial fibrillation.Cardiovasc. Ultrasound. 2014; 12: 10https://doi.org/10.1186/1476-7120-12-10
- Effect of elevated left ventricular diastolic filling pressure on the frequency of left atrial appendage thrombus in patients with nonvalvular atrial fibrillation.Am. J. Cardiol. 2011; 107: 417-422https://doi.org/10.1016/j.amjcard.2010.09.042
- Interleukin-6 and C-reactive protein and risk for death and cardiovascular events in patients with atrial fibrillation.Am. Heart J. 2015; 170: 1151-1160https://doi.org/10.1016/j.ahj.2015.09.018
- Inflammation and the pathogenesis of atrial fibrillation.Nat. Rev. Cardiol. 2015; 12: 230-243https://doi.org/10.1038/nrcardio.2015.2
- Myocardial infarction caused by surgery: blame inflammation not the surgeon.Atherosclerosis. 2016; 255: 113-116https://doi.org/10.1016/j.atherosclerosis.2016.08.044
- The sum of two evils: pneumonia and myocardial infarction: is platelet activation the missing link?.J. Am. Coll. Cardiol. 2014; 64: 1926-1928https://doi.org/10.1016/j.jacc.2014.08.023
- Inflammation in atrial fibrillation.J. Am. Coll. Cardiol. 2012; 60: 2263-2270https://doi.org/10.1016/j.jacc.2012.04.063
- Mechanisms of thrombogenesis in atrial fibrillation: Virchow's triad revisited.Lancet. 2009; 373: 155-166https://doi.org/10.1016/S0140-6736(09)60040-4
- Platelet and neutrophil responses to gram positive pathogens in patients with bacteremic infection.PLoS One. 2011; 6e26928https://doi.org/10.1371/journal.pone.0026928
- The CHADS-VASc score strongly correlates with glomerular filtration rate and predicts renal function decline over time in elderly patients with atrial fibrillation and chronic kidney disease.Int. J. Cardiol. 2018; 253: 71-77https://doi.org/10.1016/j.ijcard.2017.10.110
- Relation of renal function to risk for incident atrial fibrillation in women.Am. J. Cardiol. 2012; 109: 538-542https://doi.org/10.1016/j.amjcard.2011.10.006
- Pathophysiological mechanisms of atrial fibrillation: a translational appraisal.Physiol. Rev. 2011; 91: 265-325https://doi.org/10.1152/physrev.00031.2009
- Atrial fibrillation down-regulates renal neutral endopeptidase expression and induces profibrotic pathways in the kidney.Europace. 2008; 10: 1212-1217https://doi.org/10.1093/europace/eun206
- Anticoagulants for stroke prevention in atrial fibrillation in elderly patients.Cardiovasc. Drugs Ther. 2020; 34: 555-568https://doi.org/10.1007/s10557-020-06981-3
- Rivaroxaban: a review for secondary CV prevention in CAD and PAD.Drugs. 2020; 80: 1465-1475https://doi.org/10.1007/s40265-020-01397-7
- Antithrombotic therapy for atrial fibrillation and coronary disease demystified.Can. J. Cardiol. 2018; 34: 1426-1436https://doi.org/10.1016/j.cjca.2018.08.028
Article info
Publication history
Accepted:
January 3,
2023
Received in revised form:
November 16,
2022
Received:
October 11,
2022
Publication stage
In Press Journal Pre-ProofIdentification
Copyright
© 2023 Published by Elsevier Ltd.
