Research on left atrial appendage thrombogenic milieu prediction model in patients with nonvalvular atrial fibrillation based on machine learning algorithm

Research on left atrial appendage thrombogenic milieu prediction model in patients with nonvalvular atrial fibrillation based on machine learning algorithm | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Research on left atrial appendage thrombogenic milieu prediction model in patients with nonvalvular atrial fibrillation based on machine learning algorithm Ling Song, Xiaoqi Niu, Binbin Wang, Xiang Xu, Chen Wan, Feng Liu, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7301811/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 30 Dec, 2025 Read the published version in BMC Cardiovascular Disorders → Version 1 posted 10 You are reading this latest preprint version Abstract Objective To establish a machine-learning model of left atrial appendage thrombogenic milieu (LAATM) in patients with nonvalvular atrial fibrillation (NVAF), and analyze the corresponding risk factors to guide clinical decision-making. Methods Patients with NVAF were selected and divided into LAATM group and non-LAATM group, according to the results of transesophageal echocardiography (TEE). The LAATM group included LAA thrombus formation, sludge and spontaneous echo contrast. The patient data was collected and preprocessed. The machine learning algorithms of random forest (RF), support vector machine (SVM) and extreme gradient boosting (XGBoost) were used to establish a predictive model for LAATM in patients with NVAF. Shapley additive explanation (SHAP) was used to sort the feature importance of clinical factors. Results A total of 1217 patients were selected in this study, including 112 patients in LAATM group and 1105 patients in non-LAATM group. In terms of predictive performance, AUC value of RF model was 0.97, F1 score was 0.93, accuracy was 0.98, precision was 0.99, and recall was 0.89; AUC value of SVM model is 0.96, F1 score is 0.89, accuracy is 0.97, precision is 0.95, recall is 0.84; The AUC value of the XGBoost model is 0.96, F1 score is 0.88, accuracy is 0.97, precision is 0.98, and recall is 0.82. The prediction efficiency of RF model is the best. The prediction results of RF model were visualized by SHAP diagram, indicating that Homocysteine (HCY), NT-proBNP, C-reactive protein(CRP), glycosylated hemoglobin (HbA1c) and ABC stroke score were the top five risk factors affecting the formation of LAATM in patients with NVAF. Conclusion The RF model achieved the best predictive performance between the three prediction model. HCY, NT-proBNP, CRP, HbA1c and ABC stroke score were the top five risk factors affecting the formation of LAATM in patients with NVAF. NVAF LAATM prediction model machine learning HCY Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 1. Introduction Atrial fibrillation (AF) is a common arrhythmia in clinical practice, and ischemic stroke caused by it is an important cause of death and disability in adults, and seriously affects the quality of life of patients [ 1 ]. Non-valvular atrial fibrillation (NVAF) is the most common type of AF. Studies have shown that the thrombi in patients with NVAF is almost entirely from the left atrial appendage (LAA) [ 2 ]. The status of LAATM (LAA thrombogenic or LAA pre-thrombogenic) is closely related to adverse outcomes of patients [ 3 – 5 ]. Studies have shown that AF increased the risk of stroke and systemic embolism in patients, and the incidence of thromboembolic complications is significantly reduced after early identification of LAA thrombosis and standardized treatment[ 6 ]. Transesophageal Echocardiography (TEE) is recognized as the gold standard for the detection of LAA thrombotic status [ 7 ]. However, it has certain of the contraindications and some patients cannot tolerate the implementation process, thus its clinical application is limited to some extent. Therefore, early detection of LAA thrombosis has become an important link in preventing stroke events and improving prognosis of patients with AF. Therefore, it is very essential to establish a simple, reliable and economical model to predict NVAF factors at high risk of LAA thrombosis. For the past few years, Machine Learning has developed rapidly and has been widely used in various fields of medicine as a big data analysis method [ 8 ]. However, until now, there have been no reports on the use of machine Learning method to predict the status of LAATM in patients with NVAF. Based on this, three machine learning methods, random forest (RF), support vector machine (SVM) and extreme gradient boosting (XGBoost), were used in this study to analyze clinical data of patients with NVAF, establish a predictive model to analyze and judge the high-risk factors for the formation of LAA thrombosis, and select the best model. The model is also interpreted using SHAP diagrams to assess risk characteristics. 2. Materials and methods 2.1 General Information Patients with NVAF who were hospitalized and completed TEE in the cardiology department of our hospital from January 1, 2015 to December 31, 2020 were selected. Medical data of patients clinically diagnosed as AF related on the first page of medical records were retrieved. We extracted the data through the southwest hospital big data intelligent platform (Yidu Cloud Technology Co, LTD.). Inclusion criteria: NVAF patients with age ≥ 18 years; Transesophageal Echocardiography(TTE) and TEE were performed (clearly record whether there is LAATM status); Complete clinical data. Exclusion criteria: Combined with structural heart disease requiring surgical treatment; Presence of thrombus in other parts of the heart and incomplete clinical data. This study was reviewed and approved by the Ethics Committee of the First Affiliated Hospital of Army Military Medical University (Southwest Hospital) (Ethics number: (B) KY2021050). 2.2 TTE and TEE examination TTE was performed by an experienced sonographer using Philips IE33 color Doppler ultrasound, and the diagnosis was reviewed by a senior sonographer. TEE was performed under local anesthesia by an experienced sonographer using a Philips IE33 color Doppler ultrasound and transesophageal ultrasound probe X7-2t (frequency 2 ~ 7 MHz). The diagnosis was reviewed by a senior sonographer. In TEE, LAATM can be divided into LAA thrombogenic (Fig. 1 A) and LAA thrombogenic pre-thrombogenic states: 1) LAA sludge status (Fig. 1 B) : A dynamic, silty, or gel-like echo pattern that is not clearly fixed in the cardiac cycle but is not discrete from each other; 2) LAA spontaneous echo contrast (Fig. 1 C) : a smoke-like, turbulence-like image with weak echo intensity, no obvious fixed shape in the cardiac cycle and can be mutually discrete[ 9 ]; Fig. 1 D shows normal LAA. 2.3 Data Collection (1) Baseline characteristics: age, sex, body mass index (BMI), etc. (2) Medical history: type of AF, duration of AF, smoking history, deep vein thrombosis history, hypertension history, diabetes history, coronary heart disease history, etc. (3) According to the current commonly used methods for evaluating thromboembolism risk in AF, the CHA2DS2-VASc score [ 10 ], CHA2DS2-VASC-RAF score [ 11 ] and ABC stroke risk score[ 12 ]was included. (4) Medical laboratory data: blood routine, liver function, kidney function, Homocysteine (HCY), etc. (5) TTE and TEE examination data: the diameter and depth of LAA at 0°, 45°, 90°, 135°, and the thrombus formation of LAA at these Angle. 2.4 Data processing and machine learning 2.4.1 Statistical analysis In this retrospective study, cases with a missing value greater than 30%, such as pulmonary hypertension and warfarin use history, were not included. For binary categorical variables, such as smoking history, diabetes history and hypertension history, if the sample has this characteristic, the value is 1, otherwise it is 0. Males are assigned a value of 0 and females are assigned a value of 1. Multiple categorical variables, such as the NYHA, are assigned numbers from 1 to 4. Statistical analysis of the data was performed using R language version 4.0.2. Kolmogorov-Smirnov test was used to test the normal distribution of continuous data. The continuous data of normal distribution were represented by mean and standard, and the comparison between the two groups was tested by Independent Samples t-test. Continuous data with non-normal distributions were expressed as the median (P25,P75), and comparisons between the two groups were analyzed using either the Mann-Whitney U test or the Kruskal-Wallis test. Categorical data were expressed as n (%), and comparisons between the two groups were analyzed using Chi-square test or Fisher exact test. p < 0.05 were considered statistically significant. 2.4.2 Data preprocessing For the problem of missing data, this study adopts Multiple Imputation By Chained Equations (MICE) based on R language and uses the mice package to multiple interpolate data rows with missing values < 30%. If multiple data units and magnitude are different, Z-score method is adopted to standardize the data, and multiple sets of data are converted into unit-free Z-score scores, so as to unify the data standards, improve the comparability of data, and weaken the interpretation of data. For data balancing, we use the Over-Sampler method to solve the problem of lack of a few samples and noise interference. 2.4.3 Model creation Based on other research on binary classification problem and large sample data, we choose three standard supervised machine learning models: RF, SVM and XGBoost. The three machine learning methods are described as follows: RF is an integrated classification algorithm based on decision trees. It combines Bagging algorithm with random feature selection, uses Bootstrap method to extract training sets to train the model, and finally votes to determine the final category [ 13 ]. The final result, which has a strong ability to prevent overfitting, accurate classification, small model variance, insensitive to multivariate linearity and missing data, is suitable for high-dimensional or large-sample situations and is the one with the highest number of votes. His principle is shown in the Fig. 2 .SVM is a supervised binary classifier proposed in the 1990s based on the statistical VC dimension theory and the structural risk minimization principle[ 14 ]. His principle is shown in the Fig. 3 .With maximum generalization ability and minimum classification error rate, it can handle high dimensional data sets well, and is suitable for binary classification problems. XGBoost is an ensemble classification algorithm that integrates numerous weak classifiers to constitute a strong classifier. It is one of the best boosting algorithms based on the idea of boosting ensemble learning, adding regular terms to control the complexity of the model, which can improve the generalization ability of the model[ 15 ]. 2.4.4 Model Running environment The machine learning model runs on a Lenovo ThinkBook 14s notebook. Processor: AMD Ryzen 7 4800U with Radeon Graphics 1.80GHz; Windows 11 64-bit OS. Software environment: PyCharm Community Edition 2023.2.2 Integrated Development Environment; Python 3.8.0 programming language; Python packages such as pandas, numpy, matplotlib.pyplot, and xgboost. K-fold cross-validation method was used to divide the data into a training set (973 cases) and a test set (244 cases) according to the ratio of 8:2. The training set was used to adjust the model fitting parameters, and the test set was used to evaluate the model efficacy, while ensuring that the proportion of AF and non-AF samples in the two groups was the same. GridSearch method is used to adjust parameters to get the optimal parameters of each model to form a model. Detailed modeling process is shown in the Fig. 4 . 2.4.5 Model interpretability SHAP diagram is a visual model interpretation package developed by Python, which can intuitively reflect the impact of each independent variable (feature) on the result and the positive or negative magnitude. We used SHAP plots to explain salient features of patients, correlations between features and outcomes, and visual interpretations of individual patient instances at the population and individual levels. The horizontal coordinate in the figure represents the SHAP value, showing the influence of independent variables on the results. The ordinate indicates the importance of the independent variable, decreasing from top to bottom. To evaluate the model performance, we used the receiver operating characteristic curve (ROC), the area under the curve(AUC), accuracy, precision, recall and F1 value. Accuracy refers to the proportion of accurate quantities predicted in all samples. Precision rate refers to the proportion of the actual number of positive samples in all the predicted positive results, also known as the "accuracy rate". Recall rate refers to the proportion of samples that are actually positive results and are predicted to be positive results, also known as "recall rate". F1 value is the weighted harmonic average of accuracy rate and recall rate, and the closer F1 value is to 1, the better the prediction performance of the model. 3. Results 3.1 General characteristics A total of 1217 patients with NVAF were included in the study, including 112 patients with thrombosis and 1107 patients without thrombosis. Among them, 570 cases (46.8%) were female, and 529 cases (43.5%) were persistent AF. Comparison of clinical data between the two groups showed that blood HCY, C-reactive protein, NT-proBNP, D-dimer, glycosylated hemoglobin (HbA1c) and ABC stroke score in the LAATM group were higher than those in the non-LAATM group (P < 0.05). The basic data, test data and ultrasound data of the two groups are shown in Table 1 , Table 2 and Table 3 below. Table 1 Baseline characteristics of the study population Index Total(n = 1217) Non-LAATM group(n = 1105) LAATM group (n = 112) P value Type(%) PAF 688 (56.5) 658 (59.5) 30 (26.8) < 0.001 PeAF 529 (43.5) 447 (40.5) 82 (73.2) Duration(%) 47.30 (70.26) 46.71 (69.68) 53.20 (75.83) 0.352 Sex(%) Female 570 (46.8) 528 (47.8) 42 (37.5) 0.048 Male 647 (53.2) 577 (52.2) 70 (62.5) Age(year) 71.00 [61.00, 78.00] 70.00 [60.00, 77.00] 77.00 [69.75, 82.25] < 0.001 BMI(kg/m2) 23.74 [21.10, 26.58] 23.74 [21.11, 26.56] 23.82 [20.94, 27.31] 0.783 Past/current smoking(%) Non 741 (60.9) 687 (62.2) 54 (48.2) 0.005 Yes 476 (39.1) 418 (37.8) 58 (51.8) Alcohol consumption(%) Non 789 (64.8) 724 (65.5) 65 (58.0) 0.14 Yes 428 (35.2) 381 (34.5) 47 (42.0) VTE(%) Non 1126 (92.5) 1030 (93.2) 96 (85.7) 0.007 Yes 91 (7.5) 75 (6.8) 16 (14.3) Hypertension(%) Non 670 (55.1) 616 (55.7) 54 (48.2) 0.153 Yes 547 (44.9) 489 (44.3) 58 (51.8) Diabetes(%) Non 1033 (84.9) 946 (85.6) 87 (77.7) 0.036 Yes 184 (15.1) 159 (14.4) 25 (22.3) CHD(%) Non 535 (44.0) 498 (45.1) 37 (33.0) 0.019 Yes 682 (56.0) 607 (54.9) 75 (67.0) NYHA(%) I 447 (36.7) 429 (38.8) 18 (16.1) < 0.001 II 454 (37.3) 419 (37.9) 35 (31.2) III 301 (24.7) 247 (22.4) 54 (48.2) IV 15 (1.2) 10 (0.9) 5 (4.5) ABC 2.17 [1.54, 2.63] 2.11 [1.48, 2.53] 3.20 [2.45, 4.15] < 0.001 CHA2DS2-VASc 2.00 [1.00, 3.00] 2.00 [1.00, 3.00] 3.00 [3.00, 4.00] < 0.001 CHA2DS2-VASc- RAF 5.00 [2.00, 7.00] 4.00 [2.00, 6.00] 7.00 [5.75, 9.00] < 0.001 Abbreviations: BMI, body mass index; PAF, paroxysmal atrial fibrillation; PeAF, persistent atrial fibrillation; VTE, venous thromboembolism; CHD, coronary atherosclerotic heart disease; NYHA, New York College of Cardiology Cardiac Function Scale; ABC, ABC stroke score; CHA2DS2-VASc, CHA2DS2-VASc score; CHA2DS2-VASc-RAF, CHA2DS2-VASc-RAF score. Table 2 Hematological test data of the two groups Index total (n = 1217) No LAATM group(n = 1105) LAATM group (n = 112) P value CRP(mg/L) 43.34 [16.43, 71.26] 42.78 [16.93, 69.56] 44.69 [10.48, 105.73] 0.029 TC(umol/L) 3.99 [3.31, 4.75] 4.00 [3.32, 4.78] 3.76 [3.26, 4.56] 0.117 CysC(mg/L) 1.05 [0.87, 1.35] 1.05 [0.86, 1.35] 1.11 [0.92, 1.34] 0.542 MONO(10^9/L) 0.44 [0.33, 0.55] 0.44 [0.33, 0.55] 0.44 [0.36, 0.56] 0.235 PLT(10^9/L) 171.00 [133.00, 212.00] 171.00 [133.00, 212.00] 163.00 [132.25, 207.25] 0.594 cTnI(ug/L) 0.03 [0.01, 0.04] 0.03 [0.01, 0.04] 0.02 [0.01, 0.03] < 0.001 ALB(g/L) 38.10 [36.00, 40.50] 38.20 [36.30, 40.60] 36.65 [34.08, 39.35] < 0.001 HCY(umol/L) 14.57 [11.34, 17.61] 14.15 [11.03, 16.87] 41.47 [20.02, 60.15] < 0.001 hsCRP(mg/L) 3.24 [1.22, 8.82] 3.12 [1.15, 8.67] 4.94 [1.97, 9.42] 0.002 NT-proBNP(pg/mL) 2436.21[1161.88,3828.13] 2303.27 [1112.29, 3594.05] 7010.00 [3023.25, 11844.50] < 0.001 RDW-SD(fL) 44.90 [42.40, 47.70] 44.60 [42.20, 47.50] 46.90 [44.38, 48.95] < 0.001 Hb(g/L) 133.00 [121.00, 145.00] 133.00 [120.00, 145.00] 139.00 [129.00, 150.00] < 0.001 NEU(10^9/L) 3.49 [2.70, 4.70] 3.43 [2.66, 4.64] 3.98 [3.16, 5.27] 0.001 D-Di(mg/L(FEU)) 0.35 [0.17, 1.82] 0.33 [0.16, 1.77] 0.56 [0.24, 2.44] 0.006 Fib(g/L) 2.50 [2.11, 2.99] 2.48 [2.09, 2.97] 2.76 [2.40, 3.12] < 0.001 FDP(mg/L) 9.52 [1.50, 31.14] 9.71 [1.49, 33.31] 7.78 [1.90, 19.80] 0.024 APTT(s) 28.90 [26.50, 32.30] 28.90 [26.40, 32.20] 29.80 [26.80, 33.32] 0.358 eGFRmL/min/1.73m2) 67.79 [52.04, 84.67] 68.65 [53.10, 85.16] 57.96 [45.04, 75.16] < 0.001 HbA1c(%) 6.10 [5.70, 7.12] 6.10 [5.70, 7.20] 6.30 [5.90, 7.10] 0.008 Platelet count; CTnI, cardiac troponin I; APTT, activated partial thromboplastin time; ALB, albumin; HCY, homocysteine; hs-CRP, hypersensitive C-reactive protein; NT-proBNP, N-terminal B-type natriuretic peptide precursor; RDW-SD, red blood cell volume distribution width-standard deviation; Hb, hemoglobin; NEU, neutrophil count; D-Di: D-dimer; Fib, fibrinogen; FDP, fibrin degradation products; eGFR, estimated glomerular filtration rate; HbAlc, glycosylated hemoglobin. Table 3 Ultrasonic data indexes of the two groups Index total (n = 1217) No LAATM group(n = 1105) LAATM group (n = 112) P value PA(mm) 20.00 [20.00, 22.00] 20.00 [20.00, 22.00] 22.00 [20.00, 24.25] < 0.001 RA(mm) 39.00 [34.00, 45.00] 38.00 [33.00, 44.00] 45.00 [41.00, 48.00] < 0.001 LA(mm) 41.00 [37.00, 46.00] 40.00 [36.00, 46.00] 46.50 [43.00, 50.00] < 0.001 VS(mm) 9.80 [8.40, 11.30] 9.80 [8.40, 11.20] 10.35 [8.38, 12.03] 0.315 EF(%) 60.00 [54.00, 65.00] 60.00 [55.00, 65.00] 52.00 [40.75, 60.00] < 0.001 FS(%) 32.00 [28.00, 35.00] 32.00 [29.00, 35.00] 27.00 [20.00, 32.00] < 0.001 EDV(ml) 112.00 [97.00, 133.00] 110.00 [97.00, 131.00] 124.50 [100.00, 165.00] 0.001 ESV(ml) 66.00 [57.00, 77.00] 66.00 [57.00, 77.00] 65.00 [51.75, 75.00] 0.066 TR(m/s) 2.50 [2.30, 2.80] 2.50 [2.30, 2.80] 2.80 [2.50, 2.92] < 0.001 PAV(mm) 1.00 [1.00, 1.00] 1.00 [1.00, 1.10] 1.00 [0.90, 1.00] 0.001 WID0(mm) 18.00 [16.00, 20.00] 18.00 [16.00, 19.00] 19.00 [17.75, 21.00] < 0.001 DEP0(mm) 25.00 [23.00, 28.00] 25.00 [22.00, 27.00] 27.50 [26.00, 30.00] < 0.001 WID45(mm) 18.00 [16.00, 19.00] 17.00 [16.00, 19.00] 19.00 [18.00, 21.00] < 0.001 DEP45(mm) 25.00 [23.00, 27.00] 25.00 [22.00, 27.00] 28.00 [26.00, 30.25] < 0.001 WID90(mm) 18.00 [16.00, 20.00] 18.00 [16.00, 19.00] 19.00 [18.00, 21.00] < 0.001 DEP90(mm) 25.00 [23.00, 28.00] 25.00 [22.00, 27.00] 28.00 [26.00, 30.00] < 0.001 WID135(mm) 18.00 [16.00, 20.00] 18.00 [16.00, 20.00] 20.00 [18.00, 22.00] < 0.001 DEP135(mm) 25.00 [22.00, 27.00] 25.00 [22.00, 27.00] 27.80 [25.00, 30.00] < 0.001 Abbreviations: PA, Pulmonary artery; RA, right atrium; LA, left atrium; VS, interventricular septum; EF, ejection fraction; FS, left ventricular brachyaxis shortening rate; EDV, end-diastolic volume; ESV, end-systolic volume; TR, tricuspid regurgitation; PAV, velocity of pulmonary valve orifice; WID0, left auricle width 0°; DEP0, left auricle depth 0°; 3.2 Model Performance In this study, a total of 82 variables were incorporated into the machine learning model, and three classification models were established using RF, SVM and XGBoost. The accuracy of RF model was 0.98, precision was 0.99, recall was 0.89, and F1 score was 0.93. The accuracy of SVM model is 0.97, precision is 0.95, recall rate is 0.84 and F1 score is 0.89. The XGBoost model has an accuracy of 0.97, precision of 0.98, recall 0.82, and F1 score of 0.88. At the same time, the ROC curve and area under the curve (AUC value) of each model were obtained: RF: 0.97, support vector machines: 0.96, XGBoost: 0.96, as shown in Table 4 . ROC curves of each model are shown in Fig. 5 . On the whole, the best model is RF model. Table 4 Results of model effectiveness evaluation of each machine learning algorithms Aalgorithms Accuracy Precision Recall F1 score AUC RF 0.98 0.99 0.89 0.93 0.97 SVM 0.97 0.95 0.84 0.89 0.96 XGBoost 0.97 0.98 0.82 0.88 0.96 3.3 Model feasibility analysis According to the SHAP diagram of the RF model, the top 10 factors affecting the formation of LAA thrombosis in patients with NVAF were calculated as follows: high levels of blood HCY, high NT-proBNP value, high CRP value increased, high HbA1c, high ABC stroke score, high hsCRP value, aged, high CHA2DS2-VASc-RAF score, high D-dimer, high left atrial appendage 45° depth (Fig. 6 ). 4. Discussion Ischemic stroke is a major complication of NVAF, and LAA thrombosis is closely related to ischemic stroke, while sludge and spontaneous echo contrast are important stages of LAA thrombosis [ 16 ]. At present, more clinical studies focus on analyzing the correlation between clinical features and LAATM, while fewer studies are used to detect its potential risk. In this study, three machine learning methods, RF, SVM and XGBoost, were used to analyze the clinical data of patients with NVAF, establish a predictive model to analyze and judge the high risk factors for the formation of LAATM, and select the best model. Meanwhile, the SHAP diagram was used to interpret the model to evaluate the risk characteristics. With the application of artificial intelligence in the medical field, machine learning methods are gradually applied in medical research. In the field of cardiovascular diseases, machine learning methods are widely used in predicting prognosis and monitoring the occurrence and development process of AF, hypertension, coronary heart disease, etc. However, there are few studies using machine learning methods to predict the risk related to AF thrombosis. Zhang yiwen et al. [ 16 ]established a prediction model for thrombosis in patients with valvular atrial fibrillation by using machine learning, and pointed out that stroke volume, mitral E-wave peak velocity and tricuspid valve pressure gradient are important factors affecting thrombosis. Dong Min et al.[ 17 ] analyzed the risk factors for long-term death in elderly patients with AF combined with coronary heart disease by using multiple machine learning methods, indicated that the prediction performance of machine learning model was higher than that of traditional logistic model, and listed the top 10 factors with the highest influence factors for clinicians' reference. Jarne Verhaeghe[ 18 ] et al. established a real-time AF risk prediction model for ICU patients by using machine learning methods. Li X et al. [ 19 ] included 1864 patients with AF to conduct a study on the prediction of ischemic stroke and thromboembolism in patients. The study conducted a two-year follow-up and included 211 features for screening. Integrated machine learning methods such as generalized linear model, Bayesian model and decision tree model were used to analyze and process the data, and compared with traditional research on thrombosis prediction. The experimental results show that the predictive performance of the integrated machine learning method for AF (AUC: 0.71 ~ 0.74) is better than that of the previous thromboembolism prediction model (AUC: 0.66–0.69). Wang Xiang [ 20 ]included 60 patients with AF, screened HBG1, SNCA, GYPB, HBD, HBA2, ALAS2, SELENBP and other genes from the gene database, built a prediction model using neural network and support vector machine methods, and trained HBG1 with SNCA and GYPB neural network. It was found that the predicted data was in accordance with the original data. The SVM method was used for verification, and we found that the correlation R between the predicted value and the actual value of HBG1 by SNCA and GYPB was 0.99. The results showed that core genes HBG1, SNCA and GYPB were significantly related to AF and stroke patients who suffer from AF concurrently. In this study, the monitoring scope of ischemic stroke in NVAF was further expanded on the basis of the above studies. At the same time, machine learning methods were introduced to establish predictive models to analyze related risk factors. RF, SVM and XGBoost are commonly used machine learning models. RF is an integrated machine learning model developed on the basis of decision tree. It adopts the combination of self-service sampling and random sampling to make the model more diverse. At the same time, it can deal with missing data and data imbalance problems well, and is a nonlinear machine learning modeling tool with good performance, and has good processing ability for binary classification problems and large sample data research in this study. After the modeling process and model performance interpretation, we finally found that The accuracy of RF model was 0.98, precision was 0.99, recall was 0.89, and F1 score was 0.93, the AUC value is 0.97, and the comprehensive performance is the best. Meanwhile, HCY, NT-proBNP and C-reactive protein (CRP) were found to be independent risk factors for LAATM in patients with AF. HCY is an important intermediate product of methionine metabolism, which increases the risk of LAATM in patients with nonvalvular AF by damaging vascular endothelium [ 21 ], affecting the thrombosis characteristics of endothelial cells [ 22 ], and promoting blood hypercoagulation [ 23 ]. There are studies showing that serum HCY levels in AF patients with acute ischemic stroke are significantly higher than those in patients who have not experienced it[ 24 – 26 ]. In the baseline analysis results of this study, the median HCY in patients with LAATM was 41.47mmol/L, while the median HCY in patients without LAATM was 14.15mmol/L, with statistical significance (P < 0.05). Meanwhile, the SHAP plot results of the RF model also suggests that HCY ranks first among various influencing factors in predicting LAATM in NVAF patients.Consistent with the above research results, the increase of HCY not only causes atherosclerosis and lacunar cerebral infarction, but also is an important influencing factor in the occurrence of cardiogenic stroke. Similarly, the probability of LAATM occurrence in NVAF patients may decrease when HCY levels reduce to lower level. NT-proBNP is a non-bioactive natriuretic peptide secreted by cardiomyocytes, which is a splicing body of the same peptide as BNP, and BNP has sodium promoting, urination, vasodilation and other biological activities. Rojina Pant [ 27 ] included 261 patients with NVAF and found that BNP was correlated with LAATM. Meanwhile, multivariate logistic regression analysis demonstrated that BNP ≥ 100pg/mL was an independent predictor of LAATM. Wang Zuolan [ 28 ] included 200 patients with NVAF and divided them into thrombus group, cloud group and control group. Through the analysis of their clinical indicators, it was found that NT-proBNP was an independent high-risk factor for predicting thrombus formation in LA/LAA. The area under ROC curve for predicting thrombus formation was 0.740 (95%CI: 0.631 ~ 0.850). Multivariate logistic regression analysis showed that the OR value of NT-proBNP was 1.001.This study utilized machine learning modeling, and the results showed that NT proBNP ranked third among the factors affecting the LAATM in patients with NVAF. The baseline analysis results showed a statistically significant difference in NT-proBNP values between the LAATM group and the control group (P < 0.001), indicating that NT-proBNP has great predictive value in the thrombus formation status of the LAA in patients with NVAF, and provides a high reference value for the thrombus risk of NVAF patients with clinical heart failure. Some scholars have shown that serum CRP level is positively correlated with the risk of stroke in patients with AF, and CRP level is independently correlated with the LAA blood flow rate and TEE detection of thromboembolic risk factors. The higher the CRP level, the easier it is for TEE to detect LAA thrombosis [ 29 – 30 ]. Conway et al. found that the CRP value in patients with LA thrombus was significantly higher than that in negative patients, through a study of inflammation indicators in 37 patients with chronic AF treated with warfarin anticoagulation[ 31 ]. Many studies[ 32 ] have shown that the occurrence and development of AF can trigger inflammatory reactions, promote thrombus formation. While, inflammatory reactions are involved in the occurrence, development, and maintenance of AF. This research result further reveals the close relationship between inflammation and thrombus formation. The results of this study also suggest that CRP, an inflammatory indicator, has important value in predicting LAATM status, which is consistent with the above research results.Details are shown in Table 5 below. In this study, the scope of LAA thrombosis was expanded to include LAA thrombosis and LAA prethrombotic state (sludge and spontaneous echo contrast), and the prediction of stroke risk in patients with NVAF was more advanced, and high-risk patients could be screened for further treatment before thrombosis formed. Compared with other prediction models, more patients at high risk of stroke can be detected and further stroke events can be avoided. Moreover, we used multiple machine modeling methods to construct the model and selected the optimal model. At the same time, we revealed the interpretability of machine learning methods using SHAP graphs. 5. limitations This study also has certain limitations. This study is a single center retrospective study, and large-scale multicenter prospective studies are still needed to validate the predictive value of this model. Secondly, machine learning models have a "black box" attribute, and the correlation between their input and output data cannot be explained at present, making it difficult for us to know the specific process of their results, which still needs further verification. In future research, we should try to avoid these issues as much as possible and build models with better predictive performance. 6. Conclusion Blood HCY, NT-proBNP, CRP, HbA1c and ABC stroke score were the top five risk factors affecting the formation of LAATM in patients with NVAF. The RF model achieved the best predictive performance in the prediction model of LAATM in NVAF patients, and has high clinical application value. Declarations Acknowledgements We would like to express our gratitude to all those who helped us during the writing of this manuscript. Thanks to Department of Cardiovascular Medicine, Southwest Hospital, Army Medical University for providing data resources.Thanks to all the peer reviewers for their opinions and suggestions. The authors are also grateful for the statistical technical support of Yanxiu Chen. Authors ’ contributions HKL designed the study and is the principal investigator. LS drafted the manuscript. LS,XX,XQN and CW participated in data collection and data analysis. XQN,XZT and FL participated in manuscript writing. YW, BBW and LPL performed the statistical analysis. ZYS revised the manuscript. All authors approved the final version of the manuscript. Ethical approval This study was established, according to the ethical guidelines of the helsinki Declaration and approved by the Ethics Committee of the First Affiliated Hospital of Army Military Medical University (Southwest Hospital) (Ethics number: (B) KY2021050). All patients signed the informed consent to participate the study. Clinical trial number Not applicable. Disclosure statement No potential conffict of interest was reported by the author(s). Funding sources In preparation of this manuscript, no external funding was received. Data availability statements The data that support the findings of this study are available from the corresponding author upon reasonable request. References Joglar JA, Chung MK, Armbruster AL, Benjamin EJ, Chyou JY, Cronin EM, et al. 2023 ACC/AHA/ACCP/HRS Guideline for the Diagnosis and Management of Atrial Fibrillation:A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation. 2024;149(1):e1–156. 10.1161/CIR.0000000000001193 . Cresti A, García-Fernández MA, Sievert H, Mazzone P, Baratta P, Solari M, et al. Prevalence of extra-appendage thrombosis in non-valvular atrial fibrillation and atrial flutter in patients undergoing cardioversion:a large transoesophageal echo study. EuroIntervention. 2019;15(3):e225–30. 10.4244/EIJ-D-19-00128 . Dawn B, Varma J, Singh P, Longaker RA, Stoddard MF. Cardiovascular death in patients with atrial fibrillation is better predicted by left atrial thrombus and spontaneous echocardiographic contrast as compared with clinical parameters. J Am Soc Echocardiogr. 2005;18(3):199–205. 10.1016/j.echo.2004.12.003 . Bernhardt P, Schmidt H, Hammerstingl C, Longaker RA, Stoddard MF. Patients with atrial fibrillation and dense spontaneous echo contrast at high risk a prospective and serial follow-up over 12 months with transesophageal echocardiography and cerebral magnetic resonance imaging. J Am Coll Cardiol. 2005;45(11):1807–12. 10.1016/j.jacc. 2004.11.071 . Yoo J, Song D, Baek J-H, Kim YD, Nam HS, Hong G-R, et al. Poor Outcome of Stroke Patients With Atrial Fibrillation in the Presence of Coexisting Spontaneous Echo Contrast. Stroke. 2016;47(7):1920–2. 10.1161/STROKEAHA.116.013351 . Lurie A, Wang J, Hinnegan KJ, McIntyre WF, Belley-Côté EP, Amit G, et al. Prevalence of Left Atrial Thrombus in Anticoagulated Patients With Atrial Fibrillation. J Am Coll Cardiol. 2021;77:2875–86. 10.1016/j.jacc.2021.04.036 . Fu Y, Li K, Gao Y, Wang L, Chen M, Yang X. A novel risk score for predicting left atrial and left atrial appendage thrombogenic milieu in patients with non-valvular atrial fibrillation. Thromb Res. 2020;192:161–6. 10.1016/j.thromres.2020.05.010 . Agarwalla A, Lu Y, Reinholz AK, Marigi EM, Liu JN, Sanchez-Sotelo J. Identifying clinically meaningful subgroups following open reduction and internal fixation for proximal humerus fractures: a risk stratification analysis for mortality and 30-day complications using machine learning. JSES Int. 2024;8(5):932–40. j.jseint. 2024.04.015. He C-J, Zhang Y, Ma J-S, Liu Q, Yang J, Hou Q et al. Neutrophils/lymphocyte ratio and the state of the left auricle thrombus patients with nonvalvular atrial fibrillation. Chinese Circulation Journal. 2021;36(9):891–896. 10.3969/j.issn.1000-3614 . 2021. 09.011. Krittayaphong R, Winijkul A, Sairat P, Lip GYH. Predicting the Absolute Risk of Ischemic Stroke in Asian Patients with Atrial Fibrillation: Comparing the COOL-AF Risk Score with CARS/mCARS Models for Absolute Risk and the CHA2DS2-VASc Score. J Clin Med. 2023;12(7):2449–62. 10.3390/ jcm12072449. Sun P, Guo Z-H, Yang JG, Zhao M, Sun J, Shou Z. To evaluate the predictive value of CHA2DS2-VASc-RAF score for left atrial thrombosis in patients with nonvalvular atrial fibrillation. South China J Cardiol. 2021;27(05):547–51. 10.3969/j.issn.1007-9688.2021.05.07 . Hijazi Z, Lindbäck J, Alexander JH, Hanna M, Held C, Hylek EM, et al. The ABC (age, biomarkers, clinical history) stroke risk score:a biomarker-based risk score for predicting stroke in atrial fibrillation. Eur Heart J. 2016;37:1582–90. 10.1093/ eurheartj/ehw054. Reis FJJ, Carvalho MBL, Neves GA, Nogueira LC, Meziat-Filho N. Machine learning methods in physical therapy: A scoping review of applications in clinical context. Musculoskelet Sci Pract. 2024;74:103184. 10.1016/j.msksp.2024.103184 . Khan B, Das S, Fahim NS, Banerjee S, Khan S, Al-Sadoon MK, et al. Bayesian optimized multimodal deep hybrid learning approach for tomato leaf disease classification. Sci Rep. 2024;14(1):21525. 10.1038/s41598-024-72237-x . Agarwalla A, Lu Y, Reinholz AK, Marigi EM, Liu JN, Sanchez-Sotelo J. Identifying clinically meaningful subgroups following open reduction and internal fixation for proximal humerus fractures: a risk stratification analysis for mortality and 30-day complications using machine learning. JSES Int. 2024;8(5):932–40. j. jseint.2024. 04.015. eCollection 2024 Sep. Zhang YW, Wang ZJ, Lei YF, Tong Q, Li T, Pan F et al. Prediction and characteristic analysis of cardiac thrombosis in patients with atrial fibrillation undergoing valve disease surgery based on machine learning. Chinese Journal of Clinical Thoracic and Cardiovascular Surgery. 2022;29(09):1105–1112. 10.7507/1007-4848 . 2022 04047. Dong M, Zou T, Peng BF, Shi JY, Xu L, Pei ZW, et al. Machine learning-based prediction of long-term mortality in patients with atrial fibrillation and coronary heart disease aged 60years and over. Chin J Geriatr. 2022;41(07):804–10. 10.3760/cma.j.issn.0254-9026.2022.07.012 . Verhaeghe J, De Corte T, Sauer CM, Hendriks T, Thijssens OWM, Ongenae F, et al. Generalizable calibrated machine learning models for real-time atrial fibrillation risk prediction in ICU patients. Int J Med Inf. 2023;175:105086. 10.1016/j.ijmedinf.2023.105086 . Li X, Liu H, Du X, Zhang P, Hu G, Xie G et al. Integrated Machine Learning Approaches for Predicting Ischemic Stroke and Thromboembolism in Atrial Fibrillation. AMIA Annu Symp Proc. 2016;799–807. doi: https://pubmed.ncbi.nlm.nih.gov/28269876/ Wang X. To study the core genes of atrial fibrillation complicated with stroke and their correlation with the efficacy of factor Xa inhibitors. Peking Union Med Coll. 2023. 10.27648/d.cnki.gzxhu.2023.000194 . Steed MM, Tyagi SC. Mechanisms of cardiovascular remodeling in hyperhomocysteinemia. Antioxid Redox Signal. 2011;15(7):1927–43. 10.1089/ars.2010.3721 . Maldonado C, Soni CV, Todnem ND, Pushpakumar S, Rosenberger D, Givvimani S, et al. Hyperhomocysteinemia and sudden cardiac death: potential arrhythmogenic mechanisms. Curr Vasc Pharmacol. 2010;8(1):64–74. 10.2174/157016110790226552 . Watson T, Shantsila E, Lip GYH. Mechanisms of thrombogenesis in atrial fibrillation: Virchow's triad revisited. Lancet. 2009;373(9658):155–66. 10.1016/S0140-6736(09)60040-4 . Friedman HS. Serum homocysteine and stroke in atrial fibrillation. Ann Intern Med. 2001;134(3):253–4. 10.7326/0003-4819-134-3-200102060-00027 . YanY,Mei S-S, Li J-G, Wassouf M, D'Silva O, Kehoe RF, et al. Elevated homocysteine increases the risk of left atrial/left atrial appendage thrombus in non-valvular atrial fibrillation with low CHA2DS2-VASc score. Eur Soc Cardiol. 2017;20(7):1093–8. 10.1093/europace/eux189 . Remutula.nurbahar. The influence factors of atrial thrombosis in patients with NVAF and the comparison of anticoagulant effect. Xinjiang Med Univ. 2019.doi:CNKI:CDMD:2.1018.886364. Rojina P, Mita P, Enrique G-S, Wassouf M, D'silva O, Kehoe RF, et al. Impact of B-type natriuretic peptide level on the risk of left atrial appendage thrombus in patients with nonvalvular atrial fibrillation: a prospective study. Cardiovasc Ultrasound. 2016;14:4. 10.1186/ s12947-016-0047-6. Wang ZL. Analysis of related factors of left atrial / left atrial appendage thrombosis in non-valvular atrial fibrillation patients without heart failure. Tianjin Med Univ. 2019. doi:CNKI:CDMD:2.1018.886364. Thambidorai SK, Parakh K, Martin DO, Shah TK, Wazni O, Jasper SE et al. Relation of C-reactive protein correlates with risk of thromboembolism in patients with atrial fibrillation. Am J Cardiol. 2004;94(6):805–807. 10.1016/j.amjcard . 2004.06.011. Sun Z-J, Deng T,Peng H. Evaluation of related factors in predicting left atrial thrombus in patients with non-valvular atrial fibrillation. J Cardiovasc Pulmonary Dis. 2019;38(07):716–20. 3969 /j.issn.1007-5062.2019.07.002. Conway DSG, Buggins P, Hughes E, Lip GYH. Relation of interleukin-6, C-reactive protein, and the prothrombotic state to transesophageal echocardiographic findings in atrial fibrillation. Am J Cardiol. 2004;93(11):1368–73. 10.1016/j.amjcard.2004 . Hu Y-F, Chen Y-J, Lin Y-J, Chen S-A. Inflammation and the pathogenesis of atrial fibrillation. Nat Rev Cardiol. 2015;12(4):230–43. 10.1038/nrcardio.2015.2 . Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 30 Dec, 2025 Read the published version in BMC Cardiovascular Disorders → Version 1 posted Editorial decision: Revision requested 15 Sep, 2025 Reviews received at journal 13 Sep, 2025 Reviews received at journal 05 Sep, 2025 Reviewers agreed at journal 30 Aug, 2025 Reviewers agreed at journal 29 Aug, 2025 Reviewers invited by journal 28 Aug, 2025 Editor assigned by journal 28 Aug, 2025 Editor invited by journal 19 Aug, 2025 Submission checks completed at journal 18 Aug, 2025 First submitted to journal 18 Aug, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7301811","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":508693371,"identity":"d38bdfdc-02eb-48c9-8554-5ecda9558dd6","order_by":0,"name":"Ling Song","email":"","orcid":"","institution":"Army Medical University","correspondingAuthor":false,"prefix":"","firstName":"Ling","middleName":"","lastName":"Song","suffix":""},{"id":508693372,"identity":"fc9b5251-bd2f-4483-8c14-a00d738fead0","order_by":1,"name":"Xiaoqi Niu","email":"","orcid":"","institution":"Army Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xiaoqi","middleName":"","lastName":"Niu","suffix":""},{"id":508693373,"identity":"daa1a5c3-06ab-4c7a-adbb-04a469ccaec7","order_by":2,"name":"Binbin Wang","email":"","orcid":"","institution":"Army Medical University","correspondingAuthor":false,"prefix":"","firstName":"Binbin","middleName":"","lastName":"Wang","suffix":""},{"id":508693374,"identity":"5cf2f06e-5397-4030-98ed-f9d8ed618a2b","order_by":3,"name":"Xiang Xu","email":"","orcid":"","institution":"Army Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xiang","middleName":"","lastName":"Xu","suffix":""},{"id":508693375,"identity":"fd75673c-0d6b-4428-bd7a-3569cd0cd452","order_by":4,"name":"Chen Wan","email":"","orcid":"","institution":"Army Medical University","correspondingAuthor":false,"prefix":"","firstName":"Chen","middleName":"","lastName":"Wan","suffix":""},{"id":508693376,"identity":"e9daaced-2960-42e3-803b-781eeb45f7a7","order_by":5,"name":"Feng Liu","email":"","orcid":"","institution":"Army Medical University","correspondingAuthor":false,"prefix":"","firstName":"Feng","middleName":"","lastName":"Liu","suffix":""},{"id":508693377,"identity":"d7b59c98-e9d8-4785-b8bc-ab1653c2c16f","order_by":6,"name":"Xizhi Tang","email":"","orcid":"","institution":"Army Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xizhi","middleName":"","lastName":"Tang","suffix":""},{"id":508693378,"identity":"b6f01c41-33ec-4f88-a691-e8776380c5fd","order_by":7,"name":"Wen Yan","email":"","orcid":"","institution":"Army Medical University","correspondingAuthor":false,"prefix":"","firstName":"Wen","middleName":"","lastName":"Yan","suffix":""},{"id":508693379,"identity":"e10f715c-5a15-4c3e-bd9f-dfae773ea3e2","order_by":8,"name":"Liping Liu","email":"","orcid":"","institution":"Army Medical University","correspondingAuthor":false,"prefix":"","firstName":"Liping","middleName":"","lastName":"Liu","suffix":""},{"id":508693380,"identity":"3308e967-68de-4cb4-8025-5d56b7d1ee93","order_by":9,"name":"Zhiyuan Song","email":"","orcid":"","institution":"Army Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zhiyuan","middleName":"","lastName":"Song","suffix":""},{"id":508693381,"identity":"dfbff169-49d6-4747-b1e1-e421a5c2f3df","order_by":10,"name":"Huakang Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABA0lEQVRIiWNgGAWjYBACPmYeIHkAiJmZDz5IqIAK8+DRwgbXws6WbPDhjAERWhhgWvh5zCRnthGjhZ33mMSPM3Z58s48xsa88/5E685IYHzwto1B3hynw/jSJHtuJBcbHmYrfMy7zSB3240EZsO5bQyGOxtw+sVMgucDc+LGZubNxlAtbNK8bQwJBgdwa5H886EeqIXBTJp3DlgL+29CWqR5bhxOnM/MAvR+A8QWZgJajK1lzhxP3MAMCuRjxrnbzjxslpxzTsJwAw4t/PxnDG++OVadOL//MDAqa+Rytx1PPvjhTZmNPC5b4ABJAWMDkJAgoB4I5BsIqxkFo2AUjIIRCgDoTVnS5Tr57wAAAABJRU5ErkJggg==","orcid":"","institution":"Army Medical University","correspondingAuthor":true,"prefix":"","firstName":"Huakang","middleName":"","lastName":"Li","suffix":""}],"badges":[],"createdAt":"2025-08-05 14:38:24","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7301811/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7301811/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12872-025-05396-y","type":"published","date":"2025-12-30T15:58:07+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":90801420,"identity":"843355a9-5541-4278-8990-d81badc75c74","added_by":"auto","created_at":"2025-09-08 10:15:43","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":77347,"visible":true,"origin":"","legend":"\u003cp\u003eTEE shows the thrombus status of the left atrial appendage\u003c/p\u003e\n\u003cp\u003e(A. LAA thrombus B. LAA sludge status C. LAA spontaneous echo contrast D.Normal LAA ).\u003c/p\u003e","description":"","filename":"Figure1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7301811/v1/a23aba9db258c2d655eeca30.jpg"},{"id":90801423,"identity":"9e6020c7-74c4-4dbe-81a4-7bfb220b5355","added_by":"auto","created_at":"2025-09-08 10:15:43","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":56147,"visible":true,"origin":"","legend":"\u003cp\u003eThe principle of random forest\u003c/p\u003e","description":"","filename":"Figure2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7301811/v1/3769a8e7f511ffd730584598.jpg"},{"id":90801424,"identity":"7f868b07-d25a-4658-8097-d08a1fd52d16","added_by":"auto","created_at":"2025-09-08 10:15:43","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":79999,"visible":true,"origin":"","legend":"\u003cp\u003eThe principle of support vector machine\u003c/p\u003e","description":"","filename":"Figure3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7301811/v1/28d63cda01492345085ec255.jpg"},{"id":90801431,"identity":"cfa66013-b78e-4bf6-ba0c-0a67ceaba0c8","added_by":"auto","created_at":"2025-09-08 10:15:43","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":41528,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart of building a machine learning prediction model\u003c/p\u003e","description":"","filename":"Figure4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7301811/v1/737c5f5dc29373bb48f73f48.jpg"},{"id":90801433,"identity":"5dfba437-7e6c-473d-8dc2-52d2a46275f3","added_by":"auto","created_at":"2025-09-08 10:15:43","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":38860,"visible":true,"origin":"","legend":"\u003cp\u003eROC curves of the three models\u003c/p\u003e","description":"","filename":"Figure5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7301811/v1/64617dd4bf9d2d4f215e7535.jpg"},{"id":90803411,"identity":"6bac4955-75f1-4933-9e6f-730c1dd04e26","added_by":"auto","created_at":"2025-09-08 10:31:43","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":45798,"visible":true,"origin":"","legend":"\u003cp\u003eSHAP bar charts of 20 influencing factors derived from the RF model\u003c/p\u003e","description":"","filename":"Figure6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7301811/v1/917bccb3a3f5b3c10adc144c.jpg"},{"id":99545524,"identity":"af671cc3-f9ca-40ae-ad4e-03b5d377bf34","added_by":"auto","created_at":"2026-01-05 16:08:19","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1365784,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7301811/v1/d7bf1550-86fe-46df-9a52-1a15e8ddce66.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Research on left atrial appendage thrombogenic milieu prediction model in patients with nonvalvular atrial fibrillation based on machine learning algorithm","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eAtrial fibrillation (AF) is a common arrhythmia in clinical practice, and ischemic stroke caused by it is an important cause of death and disability in adults, and seriously affects the quality of life of patients [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Non-valvular atrial fibrillation (NVAF) is the most common type of AF. Studies have shown that the thrombi in patients with NVAF is almost entirely from the left atrial appendage (LAA) [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. The status of LAATM (LAA thrombogenic or LAA pre-thrombogenic) is closely related to adverse outcomes of patients [\u003cspan additionalcitationids=\"CR4\" citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Studies have shown that AF increased the risk of stroke and systemic embolism in patients, and the incidence of thromboembolic complications is significantly reduced after early identification of LAA thrombosis and standardized treatment[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Transesophageal Echocardiography (TEE) is recognized as the gold standard for the detection of LAA thrombotic status [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. However, it has certain of the contraindications and some patients cannot tolerate the implementation process, thus its clinical application is limited to some extent. Therefore, early detection of LAA thrombosis has become an important link in preventing stroke events and improving prognosis of patients with AF. Therefore, it is very essential to establish a simple, reliable and economical model to predict NVAF factors at high risk of LAA thrombosis.\u003c/p\u003e\u003cp\u003eFor the past few years, Machine Learning has developed rapidly and has been widely used in various fields of medicine as a big data analysis method [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. However, until now, there have been no reports on the use of machine Learning method to predict the status of LAATM in patients with NVAF. Based on this, three machine learning methods, random forest (RF), support vector machine (SVM) and extreme gradient boosting (XGBoost), were used in this study to analyze clinical data of patients with NVAF, establish a predictive model to analyze and judge the high-risk factors for the formation of LAA thrombosis, and select the best model. The model is also interpreted using SHAP diagrams to assess risk characteristics.\u003c/p\u003e"},{"header":"2. Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1 General Information\u003c/h2\u003e\u003cp\u003ePatients with NVAF who were hospitalized and completed TEE in the cardiology department of our hospital from January 1, 2015 to December 31, 2020 were selected. Medical data of patients clinically diagnosed as AF related on the first page of medical records were retrieved. We extracted the data through the southwest hospital big data intelligent platform (Yidu Cloud Technology Co, LTD.).\u003c/p\u003e\u003cp\u003eInclusion criteria: NVAF patients with age\u0026thinsp;\u0026ge;\u0026thinsp;18 years; Transesophageal Echocardiography(TTE) and TEE were performed (clearly record whether there is LAATM status); Complete clinical data. Exclusion criteria: Combined with structural heart disease requiring surgical treatment; Presence of thrombus in other parts of the heart and incomplete clinical data. This study was reviewed and approved by the Ethics Committee of the First Affiliated Hospital of Army Military Medical University (Southwest Hospital) (Ethics number: (B) KY2021050).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2 TTE and TEE examination\u003c/h2\u003e\u003cp\u003eTTE was performed by an experienced sonographer using Philips IE33 color Doppler ultrasound, and the diagnosis was reviewed by a senior sonographer. TEE was performed under local anesthesia by an experienced sonographer using a Philips IE33 color Doppler ultrasound and transesophageal ultrasound probe X7-2t (frequency 2\u0026thinsp;~\u0026thinsp;7 MHz). The diagnosis was reviewed by a senior sonographer. In TEE, LAATM can be divided into LAA thrombogenic (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA) and LAA thrombogenic pre-thrombogenic states: 1) LAA sludge status (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB) : A dynamic, silty, or gel-like echo pattern that is not clearly fixed in the cardiac cycle but is not discrete from each other; 2) LAA spontaneous echo contrast (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC) : a smoke-like, turbulence-like image with weak echo intensity, no obvious fixed shape in the cardiac cycle and can be mutually discrete[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]; Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eD shows normal LAA.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e2.3 Data Collection\u003c/h2\u003e\u003cp\u003e(1) Baseline characteristics: age, sex, body mass index (BMI), etc.\u003c/p\u003e\u003cp\u003e(2) Medical history: type of AF, duration of AF, smoking history, deep vein thrombosis history, hypertension history, diabetes history, coronary heart disease history, etc.\u003c/p\u003e\u003cp\u003e(3) According to the current commonly used methods for evaluating thromboembolism risk in AF, the CHA2DS2-VASc score [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e], CHA2DS2-VASC-RAF score [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e] and ABC stroke risk score[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]was included.\u003c/p\u003e\u003cp\u003e(4) Medical laboratory data: blood routine, liver function, kidney function, Homocysteine (HCY), etc.\u003c/p\u003e\u003cp\u003e(5) TTE and TEE examination data: the diameter and depth of LAA at 0\u0026deg;, 45\u0026deg;, 90\u0026deg;, 135\u0026deg;, and the thrombus formation of LAA at these Angle.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.4 Data processing and machine learning\u003c/h2\u003e\u003cdiv id=\"Sec7\" class=\"Section3\"\u003e\u003ch2\u003e2.4.1 Statistical analysis\u003c/h2\u003e\u003cp\u003eIn this retrospective study, cases with a missing value greater than 30%, such as pulmonary hypertension and warfarin use history, were not included. For binary categorical variables, such as smoking history, diabetes history and hypertension history, if the sample has this characteristic, the value is 1, otherwise it is 0. Males are assigned a value of 0 and females are assigned a value of 1. Multiple categorical variables, such as the NYHA, are assigned numbers from 1 to 4. Statistical analysis of the data was performed using R language version 4.0.2. Kolmogorov-Smirnov test was used to test the normal distribution of continuous data. The continuous data of normal distribution were represented by mean and standard, and the comparison between the two groups was tested by Independent Samples t-test. Continuous data with non-normal distributions were expressed as the median (P25,P75), and comparisons between the two groups were analyzed using either the Mann-Whitney U test or the Kruskal-Wallis test. Categorical data were expressed as n (%), and comparisons between the two groups were analyzed using Chi-square test or Fisher exact test. p\u0026thinsp;\u0026lt;\u0026thinsp;0.05 were considered statistically significant.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec8\" class=\"Section3\"\u003e\u003ch2\u003e2.4.2 Data preprocessing\u003c/h2\u003e\u003cp\u003eFor the problem of missing data, this study adopts Multiple Imputation By Chained Equations (MICE) based on R language and uses the mice package to multiple interpolate data rows with missing values\u0026thinsp;\u0026lt;\u0026thinsp;30%.\u003c/p\u003e\u003cp\u003eIf multiple data units and magnitude are different, Z-score method is adopted to standardize the data, and multiple sets of data are converted into unit-free Z-score scores, so as to unify the data standards, improve the comparability of data, and weaken the interpretation of data.\u003c/p\u003e\u003cp\u003eFor data balancing, we use the Over-Sampler method to solve the problem of lack of a few samples and noise interference.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec9\" class=\"Section3\"\u003e\u003ch2\u003e2.4.3 Model creation\u003c/h2\u003e\u003cp\u003eBased on other research on binary classification problem and large sample data, we choose three standard supervised machine learning models: RF, SVM and XGBoost. The three machine learning methods are described as follows:\u003c/p\u003e\u003cp\u003eRF is an integrated classification algorithm based on decision trees. It combines Bagging algorithm with random feature selection, uses Bootstrap method to extract training sets to train the model, and finally votes to determine the final category [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. The final result, which has a strong ability to prevent overfitting, accurate classification, small model variance, insensitive to multivariate linearity and missing data, is suitable for high-dimensional or large-sample situations and is the one with the highest number of votes. His principle is shown in the Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.SVM is a supervised binary classifier proposed in the 1990s based on the statistical VC dimension theory and the structural risk minimization principle[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. His principle is shown in the Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.With maximum generalization ability and minimum classification error rate, it can handle high dimensional data sets well, and is suitable for binary classification problems. XGBoost is an ensemble classification algorithm that integrates numerous weak classifiers to constitute a strong classifier. It is one of the best boosting algorithms based on the idea of boosting ensemble learning, adding regular terms to control the complexity of the model, which can improve the generalization ability of the model[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e].\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec10\" class=\"Section3\"\u003e\u003ch2\u003e2.4.4 Model Running environment\u003c/h2\u003e\u003cp\u003eThe machine learning model runs on a Lenovo ThinkBook 14s notebook. Processor: AMD Ryzen 7 4800U with Radeon Graphics 1.80GHz; Windows 11 64-bit OS. Software environment: PyCharm Community Edition 2023.2.2 Integrated Development Environment; Python 3.8.0 programming language; Python packages such as pandas, numpy, matplotlib.pyplot, and xgboost. K-fold cross-validation method was used to divide the data into a training set (973 cases) and a test set (244 cases) according to the ratio of 8:2. The training set was used to adjust the model fitting parameters, and the test set was used to evaluate the model efficacy, while ensuring that the proportion of AF and non-AF samples in the two groups was the same. GridSearch method is used to adjust parameters to get the optimal parameters of each model to form a model. Detailed modeling process is shown in the Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec11\" class=\"Section3\"\u003e\u003ch2\u003e2.4.5 Model interpretability\u003c/h2\u003e\u003cp\u003eSHAP diagram is a visual model interpretation package developed by Python, which can intuitively reflect the impact of each independent variable (feature) on the result and the positive or negative magnitude. We used SHAP plots to explain salient features of patients, correlations between features and outcomes, and visual interpretations of individual patient instances at the population and individual levels. The horizontal coordinate in the figure represents the SHAP value, showing the influence of independent variables on the results. The ordinate indicates the importance of the independent variable, decreasing from top to bottom.\u003c/p\u003e\u003cp\u003eTo evaluate the model performance, we used the receiver operating characteristic curve (ROC), the area under the curve(AUC), accuracy, precision, recall and F1 value. Accuracy refers to the proportion of accurate quantities predicted in all samples. Precision rate refers to the proportion of the actual number of positive samples in all the predicted positive results, also known as the \"accuracy rate\". Recall rate refers to the proportion of samples that are actually positive results and are predicted to be positive results, also known as \"recall rate\". F1 value is the weighted harmonic average of accuracy rate and recall rate, and the closer F1 value is to 1, the better the prediction performance of the model.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003e3.1 General characteristics\u003c/h2\u003e\u003cp\u003eA total of 1217 patients with NVAF were included in the study, including 112 patients with thrombosis and 1107 patients without thrombosis. Among them, 570 cases (46.8%) were female, and 529 cases (43.5%) were persistent AF. Comparison of clinical data between the two groups showed that blood HCY, C-reactive protein, NT-proBNP, D-dimer, glycosylated hemoglobin (HbA1c) and ABC stroke score in the LAATM group were higher than those in the non-LAATM group (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05). The basic data, test data and ultrasound data of the two groups are shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e below.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eBaseline characteristics of the study population \u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIndex\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eTotal(n\u0026thinsp;=\u0026thinsp;1217)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eNon-LAATM group(n\u0026thinsp;=\u0026thinsp;1105)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eLAATM group (n\u0026thinsp;=\u0026thinsp;112)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eP value\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eType(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePAF\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e688 (56.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e658 (59.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e30 (26.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePeAF\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e529 (43.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e447 (40.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e82 (73.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDuration(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e47.30 (70.26)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e46.71 (69.68)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e53.20 (75.83)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.352\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSex(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eFemale\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e570 (46.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e528 (47.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e42 (37.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.048\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMale\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e647 (53.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e577 (52.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e70 (62.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge(year)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e71.00 [61.00, 78.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e70.00 [60.00, 77.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e77.00 [69.75, 82.25]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBMI(kg/m2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e23.74 [21.10, 26.58]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e23.74 [21.11, 26.56]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e23.82 [20.94, 27.31]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.783\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePast/current smoking(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNon\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e741 (60.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e687 (62.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e54 (48.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.005\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e476 (39.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e418 (37.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e58 (51.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAlcohol consumption(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNon\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e789 (64.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e724 (65.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e65 (58.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.14\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e428 (35.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e381 (34.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e47 (42.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVTE(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNon\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1126 (92.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1030 (93.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e96 (85.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.007\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e91 (7.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e75 (6.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e16 (14.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHypertension(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNon\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e670 (55.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e616 (55.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e54 (48.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.153\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e547 (44.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e489 (44.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e58 (51.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDiabetes(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNon\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1033 (84.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e946 (85.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e87 (77.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.036\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e184 (15.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e159 (14.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e25 (22.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCHD(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNon\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e535 (44.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e498 (45.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e37 (33.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.019\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e682 (56.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e607 (54.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e75 (67.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNYHA(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eI\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e447 (36.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e429 (38.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e18 (16.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eII\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e454 (37.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e419 (37.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e35 (31.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eIII\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e301 (24.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e247 (22.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e54 (48.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eIV\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e15 (1.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e10 (0.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e5 (4.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eABC\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e2.17 [1.54, 2.63]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e2.11 [1.48, 2.53]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e3.20 [2.45, 4.15]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCHA2DS2-VASc\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e2.00 [1.00, 3.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e2.00 [1.00, 3.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e3.00 [3.00, 4.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCHA2DS2-VASc- RAF\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e5.00 [2.00, 7.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e4.00 [2.00, 6.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e7.00 [5.75, 9.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eAbbreviations: BMI, body mass index; PAF, paroxysmal atrial fibrillation; PeAF, persistent atrial fibrillation; VTE, venous thromboembolism; CHD, coronary atherosclerotic heart disease; NYHA, New York College of Cardiology Cardiac Function Scale; ABC, ABC stroke score; CHA2DS2-VASc, CHA2DS2-VASc score; CHA2DS2-VASc-RAF, CHA2DS2-VASc-RAF score.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eHematological test data of the two groups\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIndex\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003etotal (n\u0026thinsp;=\u0026thinsp;1217)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eNo LAATM group(n\u0026thinsp;=\u0026thinsp;1105)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eLAATM group (n\u0026thinsp;=\u0026thinsp;112)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eP value\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCRP(mg/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e43.34 [16.43, 71.26]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e42.78 [16.93, 69.56]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e44.69 [10.48, 105.73]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.029\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTC(umol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e3.99 [3.31, 4.75]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e4.00 [3.32, 4.78]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e3.76 [3.26, 4.56]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.117\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCysC(mg/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1.05 [0.87, 1.35]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.05 [0.86, 1.35]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e1.11 [0.92, 1.34]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.542\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMONO(10^9/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.44 [0.33, 0.55]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.44 [0.33, 0.55]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.44 [0.36, 0.56]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.235\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePLT(10^9/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e171.00 [133.00, 212.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e171.00 [133.00, 212.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e163.00 [132.25, 207.25]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.594\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ecTnI(ug/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.03 [0.01, 0.04]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.03 [0.01, 0.04]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.02 [0.01, 0.03]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALB(g/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e38.10 [36.00, 40.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e38.20 [36.30, 40.60]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e36.65 [34.08, 39.35]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHCY(umol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e14.57 [11.34, 17.61]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e14.15 [11.03, 16.87]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e41.47 [20.02, 60.15]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ehsCRP(mg/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e3.24 [1.22, 8.82]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e3.12 [1.15, 8.67]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e4.94 [1.97, 9.42]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.002\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNT-proBNP(pg/mL)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e2436.21[1161.88,3828.13]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e2303.27 [1112.29, 3594.05]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e7010.00 [3023.25, 11844.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRDW-SD(fL)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e44.90 [42.40, 47.70]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e44.60 [42.20, 47.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e46.90 [44.38, 48.95]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHb(g/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e133.00 [121.00, 145.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e133.00 [120.00, 145.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e139.00 [129.00, 150.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNEU(10^9/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e3.49 [2.70, 4.70]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e3.43 [2.66, 4.64]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e3.98 [3.16, 5.27]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eD-Di(mg/L(FEU))\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.35 [0.17, 1.82]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.33 [0.16, 1.77]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.56 [0.24, 2.44]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.006\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFib(g/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e2.50 [2.11, 2.99]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e2.48 [2.09, 2.97]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e2.76 [2.40, 3.12]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFDP(mg/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e9.52 [1.50, 31.14]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e9.71 [1.49, 33.31]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e7.78 [1.90, 19.80]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.024\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAPTT(s)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e28.90 [26.50, 32.30]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e28.90 [26.40, 32.20]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e29.80 [26.80, 33.32]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.358\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eeGFRmL/min/1.73m2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e67.79 [52.04, 84.67]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e68.65 [53.10, 85.16]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e57.96 [45.04, 75.16]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHbA1c(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e6.10 [5.70, 7.12]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e6.10 [5.70, 7.20]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e6.30 [5.90, 7.10]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.008\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"6\"\u003ePlatelet count; CTnI, cardiac troponin I; APTT, activated partial thromboplastin time; ALB, albumin; HCY, homocysteine; hs-CRP, hypersensitive C-reactive protein; NT-proBNP, N-terminal B-type natriuretic peptide precursor; RDW-SD, red blood cell volume distribution width-standard deviation; Hb, hemoglobin; NEU, neutrophil count; D-Di: D-dimer; Fib, fibrinogen; FDP, fibrin degradation products; eGFR, estimated glomerular filtration rate; HbAlc, glycosylated hemoglobin.\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eUltrasonic data indexes of the two groups\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIndex\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003etotal (n\u0026thinsp;=\u0026thinsp;1217)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNo LAATM group(n\u0026thinsp;=\u0026thinsp;1105)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eLAATM group (n\u0026thinsp;=\u0026thinsp;112)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eP value\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePA(mm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e20.00 [20.00, 22.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e20.00 [20.00, 22.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e22.00 [20.00, 24.25]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRA(mm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e39.00 [34.00, 45.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e38.00 [33.00, 44.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e45.00 [41.00, 48.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLA(mm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e41.00 [37.00, 46.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e40.00 [36.00, 46.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e46.50 [43.00, 50.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVS(mm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e9.80 [8.40, 11.30]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e9.80 [8.40, 11.20]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e10.35 [8.38, 12.03]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.315\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEF(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e60.00 [54.00, 65.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e60.00 [55.00, 65.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e52.00 [40.75, 60.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFS(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e32.00 [28.00, 35.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e32.00 [29.00, 35.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e27.00 [20.00, 32.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEDV(ml)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e112.00 [97.00, 133.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e110.00 [97.00, 131.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e124.50 [100.00, 165.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eESV(ml)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e66.00 [57.00, 77.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e66.00 [57.00, 77.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e65.00 [51.75, 75.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.066\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTR(m/s)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e2.50 [2.30, 2.80]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e2.50 [2.30, 2.80]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e2.80 [2.50, 2.92]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePAV(mm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.00 [1.00, 1.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1.00 [1.00, 1.10]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.00 [0.90, 1.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWID0(mm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e18.00 [16.00, 20.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e18.00 [16.00, 19.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e19.00 [17.75, 21.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDEP0(mm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e25.00 [23.00, 28.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e25.00 [22.00, 27.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e27.50 [26.00, 30.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWID45(mm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e18.00 [16.00, 19.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e17.00 [16.00, 19.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e19.00 [18.00, 21.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDEP45(mm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e25.00 [23.00, 27.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e25.00 [22.00, 27.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e28.00 [26.00, 30.25]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWID90(mm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e18.00 [16.00, 20.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e18.00 [16.00, 19.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e19.00 [18.00, 21.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDEP90(mm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e25.00 [23.00, 28.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e25.00 [22.00, 27.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e28.00 [26.00, 30.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWID135(mm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e18.00 [16.00, 20.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e18.00 [16.00, 20.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e20.00 [18.00, 22.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDEP135(mm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e25.00 [22.00, 27.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e25.00 [22.00, 27.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e27.80 [25.00, 30.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"5\"\u003eAbbreviations: PA, Pulmonary artery; RA, right atrium; LA, left atrium; VS, interventricular septum; EF, ejection fraction; FS, left ventricular brachyaxis shortening rate; EDV, end-diastolic volume; ESV, end-systolic volume; TR, tricuspid regurgitation; PAV, velocity of pulmonary valve orifice; WID0, left auricle width 0\u0026deg;; DEP0, left auricle depth 0\u0026deg;;\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003e3.2 Model Performance\u003c/h2\u003e\u003cp\u003eIn this study, a total of 82 variables were incorporated into the machine learning model, and three classification models were established using RF, SVM and XGBoost. The accuracy of RF model was 0.98, precision was 0.99, recall was 0.89, and F1 score was 0.93. The accuracy of SVM model is 0.97, precision is 0.95, recall rate is 0.84 and F1 score is 0.89. The XGBoost model has an accuracy of 0.97, precision of 0.98, recall 0.82, and F1 score of 0.88.\u003c/p\u003e\u003cp\u003eAt the same time, the ROC curve and area under the curve (AUC value) of each model were obtained: RF: 0.97, support vector machines: 0.96, XGBoost: 0.96, as shown in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e. ROC curves of each model are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e. On the whole, the best model is RF model.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eResults of model effectiveness evaluation of each machine learning algorithms\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"7\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAalgorithms\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eAccuracy\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePrecision\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eRecall\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e\u003cp\u003eF1 score\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e\u003cp\u003eAUC\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRF\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.98\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.99\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.89\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.93\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e\u003cp\u003e0.97\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSVM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.97\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.95\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.84\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.89\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e\u003cp\u003e0.96\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eXGBoost\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.97\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.98\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.82\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.88\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e\u003cp\u003e0.96\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003e3.3 Model feasibility analysis\u003c/h2\u003e\u003cp\u003eAccording to the SHAP diagram of the RF model, the top 10 factors affecting the formation of LAA thrombosis in patients with NVAF were calculated as follows: high levels of blood HCY, high NT-proBNP value, high CRP value increased, high HbA1c, high ABC stroke score, high hsCRP value, aged, high CHA2DS2-VASc-RAF score, high D-dimer, high left atrial appendage 45\u0026deg; depth (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eIschemic stroke is a major complication of NVAF, and LAA thrombosis is closely related to ischemic stroke, while sludge and spontaneous echo contrast are important stages of LAA thrombosis [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. At present, more clinical studies focus on analyzing the correlation between clinical features and LAATM, while fewer studies are used to detect its potential risk. In this study, three machine learning methods, RF, SVM and XGBoost, were used to analyze the clinical data of patients with NVAF, establish a predictive model to analyze and judge the high risk factors for the formation of LAATM, and select the best model. Meanwhile, the SHAP diagram was used to interpret the model to evaluate the risk characteristics.\u003c/p\u003e\u003cp\u003eWith the application of artificial intelligence in the medical field, machine learning methods are gradually applied in medical research. In the field of cardiovascular diseases, machine learning methods are widely used in predicting prognosis and monitoring the occurrence and development process of AF, hypertension, coronary heart disease, etc. However, there are few studies using machine learning methods to predict the risk related to AF thrombosis. Zhang yiwen et al. [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]established a prediction model for thrombosis in patients with valvular atrial fibrillation by using machine learning, and pointed out that stroke volume, mitral E-wave peak velocity and tricuspid valve pressure gradient are important factors affecting thrombosis. Dong Min et al.[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e] analyzed the risk factors for long-term death in elderly patients with AF combined with coronary heart disease by using multiple machine learning methods, indicated that the prediction performance of machine learning model was higher than that of traditional logistic model, and listed the top 10 factors with the highest influence factors for clinicians' reference. Jarne Verhaeghe[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e] et al. established a real-time AF risk prediction model for ICU patients by using machine learning methods. Li X et al. [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e] included 1864 patients with AF to conduct a study on the prediction of ischemic stroke and thromboembolism in patients. The study conducted a two-year follow-up and included 211 features for screening. Integrated machine learning methods such as generalized linear model, Bayesian model and decision tree model were used to analyze and process the data, and compared with traditional research on thrombosis prediction. The experimental results show that the predictive performance of the integrated machine learning method for AF (AUC: 0.71\u0026thinsp;~\u0026thinsp;0.74) is better than that of the previous thromboembolism prediction model (AUC: 0.66\u0026ndash;0.69). Wang Xiang [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]included 60 patients with AF, screened HBG1, SNCA, GYPB, HBD, HBA2, ALAS2, SELENBP and other genes from the gene database, built a prediction model using neural network and support vector machine methods, and trained HBG1 with SNCA and GYPB neural network. It was found that the predicted data was in accordance with the original data. The SVM method was used for verification, and we found that the correlation R between the predicted value and the actual value of HBG1 by SNCA and GYPB was 0.99. The results showed that core genes HBG1, SNCA and GYPB were significantly related to AF and stroke patients who suffer from AF concurrently. In this study, the monitoring scope of ischemic stroke in NVAF was further expanded on the basis of the above studies. At the same time, machine learning methods were introduced to establish predictive models to analyze related risk factors.\u003c/p\u003e\u003cp\u003eRF, SVM and XGBoost are commonly used machine learning models. RF is an integrated machine learning model developed on the basis of decision tree. It adopts the combination of self-service sampling and random sampling to make the model more diverse. At the same time, it can deal with missing data and data imbalance problems well, and is a nonlinear machine learning modeling tool with good performance, and has good processing ability for binary classification problems and large sample data research in this study. After the modeling process and model performance interpretation, we finally found that The accuracy of RF model was 0.98, precision was 0.99, recall was 0.89, and F1 score was 0.93, the AUC value is 0.97, and the comprehensive performance is the best. Meanwhile, HCY, NT-proBNP and C-reactive protein (CRP) were found to be independent risk factors for LAATM in patients with AF.\u003c/p\u003e\u003cp\u003eHCY is an important intermediate product of methionine metabolism, which increases the risk of LAATM in patients with nonvalvular AF by damaging vascular endothelium [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e], affecting the thrombosis characteristics of endothelial cells [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e], and promoting blood hypercoagulation [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. There are studies showing that serum HCY levels in AF patients with acute ischemic stroke are significantly higher than those in patients who have not experienced it[\u003cspan additionalcitationids=\"CR25\" citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. In the baseline analysis results of this study, the median HCY in patients with LAATM was 41.47mmol/L, while the median HCY in patients without LAATM was 14.15mmol/L, with statistical significance (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Meanwhile, the SHAP plot results of the RF model also suggests that HCY ranks first among various influencing factors in predicting LAATM in NVAF patients.Consistent with the above research results, the increase of HCY not only causes atherosclerosis and lacunar cerebral infarction, but also is an important influencing factor in the occurrence of cardiogenic stroke. Similarly, the probability of LAATM occurrence in NVAF patients may decrease when HCY levels reduce to lower level.\u003c/p\u003e\u003cp\u003eNT-proBNP is a non-bioactive natriuretic peptide secreted by cardiomyocytes, which is a splicing body of the same peptide as BNP, and BNP has sodium promoting, urination, vasodilation and other biological activities. Rojina Pant [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e] included 261 patients with NVAF and found that BNP was correlated with LAATM. Meanwhile, multivariate logistic regression analysis demonstrated that BNP\u0026thinsp;\u0026ge;\u0026thinsp;100pg/mL was an independent predictor of LAATM. Wang Zuolan [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e] included 200 patients with NVAF and divided them into thrombus group, cloud group and control group. Through the analysis of their clinical indicators, it was found that NT-proBNP was an independent high-risk factor for predicting thrombus formation in LA/LAA. The area under ROC curve for predicting thrombus formation was 0.740 (95%CI: 0.631\u0026thinsp;~\u0026thinsp;0.850). Multivariate logistic regression analysis showed that the OR value of NT-proBNP was 1.001.This study utilized machine learning modeling, and the results showed that NT proBNP ranked third among the factors affecting the LAATM in patients with NVAF. The baseline analysis results showed a statistically significant difference in NT-proBNP values between the LAATM group and the control group (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), indicating that NT-proBNP has great predictive value in the thrombus formation status of the LAA in patients with NVAF, and provides a high reference value for the thrombus risk of NVAF patients with clinical heart failure.\u003c/p\u003e\u003cp\u003eSome scholars have shown that serum CRP level is positively correlated with the risk of stroke in patients with AF, and CRP level is independently correlated with the LAA blood flow rate and TEE detection of thromboembolic risk factors. The higher the CRP level, the easier it is for TEE to detect LAA thrombosis [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. Conway et al. found that the CRP value in patients with LA thrombus was significantly higher than that in negative patients, through a study of inflammation indicators in 37 patients with chronic AF treated with warfarin anticoagulation[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. Many studies[\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e32\u003c/span\u003e] have shown that the occurrence and development of AF can trigger inflammatory reactions, promote thrombus formation. While, inflammatory reactions are involved in the occurrence, development, and maintenance of AF. This research result further reveals the close relationship between inflammation and thrombus formation. The results of this study also suggest that CRP, an inflammatory indicator, has important value in predicting LAATM status, which is consistent with the above research results.Details are shown in Table\u0026nbsp;5 below.\u003c/p\u003e\u003cp\u003eIn this study, the scope of LAA thrombosis was expanded to include LAA thrombosis and LAA prethrombotic state (sludge and spontaneous echo contrast), and the prediction of stroke risk in patients with NVAF was more advanced, and high-risk patients could be screened for further treatment before thrombosis formed. Compared with other prediction models, more patients at high risk of stroke can be detected and further stroke events can be avoided. Moreover, we used multiple machine modeling methods to construct the model and selected the optimal model. At the same time, we revealed the interpretability of machine learning methods using SHAP graphs.\u003c/p\u003e"},{"header":"5. limitations","content":"\u003cp\u003eThis study also has certain limitations. This study is a single center retrospective study, and large-scale multicenter prospective studies are still needed to validate the predictive value of this model. Secondly, machine learning models have a \"black box\" attribute, and the correlation between their input and output data cannot be explained at present, making it difficult for us to know the specific process of their results, which still needs further verification. In future research, we should try to avoid these issues as much as possible and build models with better predictive performance.\u003c/p\u003e"},{"header":"6. Conclusion","content":"\u003cp\u003eBlood HCY, NT-proBNP, CRP, HbA1c and ABC stroke score were the top five risk factors affecting the formation of LAATM in patients with NVAF. The RF model achieved the best predictive performance in the prediction model of LAATM in NVAF patients, and has high clinical application value.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe would like to express our gratitude to all those who helped us during the writing of this manuscript. Thanks to Department of Cardiovascular Medicine, Southwest Hospital, Army Medical University for providing data resources.Thanks to all the peer reviewers for their opinions and suggestions. The authors are also grateful for the statistical technical support of Yanxiu Chen.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u003c/strong\u003e\u003cstrong\u003e\u0026rsquo; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHKL designed the study and is the principal investigator. LS drafted the manuscript. LS,XX,XQN and CW participated in data collection and data analysis. XQN,XZT and FL participated in manuscript writing. YW, BBW and LPL performed the statistical analysis. ZYS revised the manuscript. All authors approved the final version of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthical approval\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was established, according to the ethical guidelines of the helsinki Declaration and approved by the Ethics Committee of the First Affiliated Hospital of Army Military Medical University (Southwest Hospital) (Ethics number: (B) KY2021050). All patients signed the informed consent to participate the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical trial number\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDisclosure statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNo potential conffict of interest was reported by the author(s).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding sources\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn preparation of this manuscript, no external funding was received.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability statements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data that support the findings of this study are available from the corresponding author upon reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003e\u003cspan\u003eJoglar JA, Chung MK, Armbruster AL, Benjamin EJ, Chyou JY, Cronin EM, et al. 2023 ACC/AHA/ACCP/HRS Guideline for the Diagnosis and Management of Atrial Fibrillation:A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation. 2024;149(1):e1\u0026ndash;156. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1161/CIR.0000000000001193\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eCresti A, Garc\u0026iacute;a-Fern\u0026aacute;ndez MA, Sievert H, Mazzone P, Baratta P, Solari M, et al. Prevalence of extra-appendage thrombosis in non-valvular atrial fibrillation and atrial flutter in patients undergoing cardioversion:a large transoesophageal echo study. EuroIntervention. 2019;15(3):e225\u0026ndash;30. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.4244/EIJ-D-19-00128\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eDawn B, Varma J, Singh P, Longaker RA, Stoddard MF. Cardiovascular death in patients with atrial fibrillation is better predicted by left atrial thrombus and spontaneous echocardiographic contrast as compared with clinical parameters. J Am Soc Echocardiogr. 2005;18(3):199\u0026ndash;205. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.echo.2004.12.003\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eBernhardt P, Schmidt H, Hammerstingl C, Longaker RA, Stoddard MF. Patients with atrial fibrillation and dense spontaneous echo contrast at high risk a prospective and serial follow-up over 12 months with transesophageal echocardiography and cerebral magnetic resonance imaging. J Am Coll Cardiol. 2005;45(11):1807\u0026ndash;12. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jacc. 2004.11.071\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eYoo J, Song D, Baek J-H, Kim YD, Nam HS, Hong G-R, et al. Poor Outcome of Stroke Patients With Atrial Fibrillation in the Presence of Coexisting Spontaneous Echo Contrast. Stroke. 2016;47(7):1920\u0026ndash;2. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1161/STROKEAHA.116.013351\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eLurie A, Wang J, Hinnegan KJ, McIntyre WF, Belley-C\u0026ocirc;t\u0026eacute; EP, Amit G, et al. Prevalence of Left Atrial Thrombus in Anticoagulated Patients With Atrial Fibrillation. J Am Coll Cardiol. 2021;77:2875\u0026ndash;86. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jacc.2021.04.036\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eFu Y, Li K, Gao Y, Wang L, Chen M, Yang X. A novel risk score for predicting left atrial and left atrial appendage thrombogenic milieu in patients with non-valvular atrial fibrillation. Thromb Res. 2020;192:161\u0026ndash;6. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.thromres.2020.05.010\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eAgarwalla A, Lu Y, Reinholz AK, Marigi EM, Liu JN, Sanchez-Sotelo J. Identifying clinically meaningful subgroups following open reduction and internal fixation for proximal humerus fractures: a risk stratification analysis for mortality and 30-day complications using machine learning. JSES Int. 2024;8(5):932\u0026ndash;40. j.jseint. 2024.04.015.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eHe C-J, Zhang Y, Ma J-S, Liu Q, Yang J, Hou Q et al. Neutrophils/lymphocyte ratio and the state of the left auricle thrombus patients with nonvalvular atrial fibrillation. Chinese Circulation Journal. 2021;36(9):891\u0026ndash;896. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3969/j.issn.1000-3614\u003c/span\u003e\u003c/span\u003e. 2021. 09.011.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eKrittayaphong R, Winijkul A, Sairat P, Lip GYH. Predicting the Absolute Risk of Ischemic Stroke in Asian Patients with Atrial Fibrillation: Comparing the COOL-AF Risk Score with CARS/mCARS Models for Absolute Risk and the CHA2DS2-VASc Score. J Clin Med. 2023;12(7):2449\u0026ndash;62. 10.3390/ jcm12072449.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eSun P, Guo Z-H, Yang JG, Zhao M, Sun J, Shou Z. To evaluate the predictive value of CHA2DS2-VASc-RAF score for left atrial thrombosis in patients with nonvalvular atrial fibrillation. South China J Cardiol. 2021;27(05):547\u0026ndash;51. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3969/j.issn.1007-9688.2021.05.07\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eHijazi Z, Lindb\u0026auml;ck J, Alexander JH, Hanna M, Held C, Hylek EM, et al. The ABC (age, biomarkers, clinical history) stroke risk score:a biomarker-based risk score for predicting stroke in atrial fibrillation. Eur Heart J. 2016;37:1582\u0026ndash;90. 10.1093/ eurheartj/ehw054.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eReis FJJ, Carvalho MBL, Neves GA, Nogueira LC, Meziat-Filho N. Machine learning methods in physical therapy: A scoping review of applications in clinical context. Musculoskelet Sci Pract. 2024;74:103184. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.msksp.2024.103184\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eKhan B, Das S, Fahim NS, Banerjee S, Khan S, Al-Sadoon MK, et al. Bayesian optimized multimodal deep hybrid learning approach for tomato leaf disease classification. Sci Rep. 2024;14(1):21525. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41598-024-72237-x\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eAgarwalla A, Lu Y, Reinholz AK, Marigi EM, Liu JN, Sanchez-Sotelo J. Identifying clinically meaningful subgroups following open reduction and internal fixation for proximal humerus fractures: a risk stratification analysis for mortality and 30-day complications using machine learning. JSES Int. 2024;8(5):932\u0026ndash;40. j. jseint.2024. 04.015. eCollection 2024 Sep.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eZhang YW, Wang ZJ, Lei YF, Tong Q, Li T, Pan F et al. Prediction and characteristic analysis of cardiac thrombosis in patients with atrial fibrillation undergoing valve disease surgery based on machine learning. Chinese Journal of Clinical Thoracic and Cardiovascular Surgery. 2022;29(09):1105\u0026ndash;1112. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.7507/1007-4848\u003c/span\u003e\u003c/span\u003e. 2022 04047.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eDong M, Zou T, Peng BF, Shi JY, Xu L, Pei ZW, et al. Machine learning-based prediction of long-term mortality in patients with atrial fibrillation and coronary heart disease aged 60years and over. Chin J Geriatr. 2022;41(07):804\u0026ndash;10. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3760/cma.j.issn.0254-9026.2022.07.012\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eVerhaeghe J, De Corte T, Sauer CM, Hendriks T, Thijssens OWM, Ongenae F, et al. Generalizable calibrated machine learning models for real-time atrial fibrillation risk prediction in ICU patients. Int J Med Inf. 2023;175:105086. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.ijmedinf.2023.105086\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eLi X, Liu H, Du X, Zhang P, Hu G, Xie G et al. Integrated Machine Learning Approaches for Predicting Ischemic Stroke and Thromboembolism in Atrial Fibrillation. AMIA Annu Symp Proc. 2016;799\u0026ndash;807. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/28269876/\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWang X. To study the core genes of atrial fibrillation complicated with stroke and their correlation with the efficacy of factor Xa inhibitors. Peking Union Med Coll. 2023. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.27648/d.cnki.gzxhu.2023.000194\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eSteed MM, Tyagi SC. Mechanisms of cardiovascular remodeling in hyperhomocysteinemia. Antioxid Redox Signal. 2011;15(7):1927\u0026ndash;43. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1089/ars.2010.3721\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eMaldonado C, Soni CV, Todnem ND, Pushpakumar S, Rosenberger D, Givvimani S, et al. Hyperhomocysteinemia and sudden cardiac death: potential arrhythmogenic mechanisms. Curr Vasc Pharmacol. 2010;8(1):64\u0026ndash;74. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2174/157016110790226552\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWatson T, Shantsila E, Lip GYH. Mechanisms of thrombogenesis in atrial fibrillation: Virchow\u0026apos;s triad revisited. Lancet. 2009;373(9658):155\u0026ndash;66. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S0140-6736(09)60040-4\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eFriedman HS. Serum homocysteine and stroke in atrial fibrillation. Ann Intern Med. 2001;134(3):253\u0026ndash;4. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.7326/0003-4819-134-3-200102060-00027\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eYanY,Mei S-S, Li J-G, Wassouf M, D\u0026apos;Silva O, Kehoe RF, et al. Elevated homocysteine increases the risk of left atrial/left atrial appendage thrombus in non-valvular atrial fibrillation with low CHA2DS2-VASc score. Eur Soc Cardiol. 2017;20(7):1093\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/europace/eux189\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eRemutula.nurbahar. The influence factors of atrial thrombosis in patients with NVAF and the comparison of anticoagulant effect. Xinjiang Med Univ. 2019.doi:CNKI:CDMD:2.1018.886364.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eRojina P, Mita P, Enrique G-S, Wassouf M, D\u0026apos;silva O, Kehoe RF, et al. Impact of B-type natriuretic peptide level on the risk of left atrial appendage thrombus in patients with nonvalvular atrial fibrillation: a prospective study. Cardiovasc Ultrasound. 2016;14:4. 10.1186/ s12947-016-0047-6.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWang ZL. Analysis of related factors of left atrial / left atrial appendage thrombosis in non-valvular atrial fibrillation patients without heart failure. Tianjin Med Univ. 2019. doi:CNKI:CDMD:2.1018.886364.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eThambidorai SK, Parakh K, Martin DO, Shah TK, Wazni O, Jasper SE et al. Relation of C-reactive protein correlates with risk of thromboembolism in patients with atrial fibrillation. Am J Cardiol. 2004;94(6):805\u0026ndash;807. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.amjcard\u003c/span\u003e\u003c/span\u003e. 2004.06.011.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eSun Z-J, Deng T,Peng H. Evaluation of related factors in predicting left atrial thrombus in patients with non-valvular atrial fibrillation. J Cardiovasc Pulmonary Dis. 2019;38(07):716\u0026ndash;20.\u003c/span\u003e\u003cspan\u003e3969 /j.issn.1007-5062.2019.07.002.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eConway DSG, Buggins P, Hughes E, Lip GYH. Relation of interleukin-6, C-reactive protein, and the prothrombotic state to transesophageal echocardiographic findings in atrial fibrillation. Am J Cardiol. 2004;93(11):1368\u0026ndash;73. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.amjcard.2004\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eHu Y-F, Chen Y-J, Lin Y-J, Chen S-A. Inflammation and the pathogenesis of atrial fibrillation. Nat Rev Cardiol. 2015;12(4):230\u0026ndash;43. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nrcardio.2015.2\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-cardiovascular-disorders","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcar","sideBox":"Learn more about [BMC Cardiovascular Disorders](http://bmccardiovascdisord.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcar/default.aspx","title":"BMC Cardiovascular Disorders","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"NVAF, LAATM, prediction model, machine learning, HCY","lastPublishedDoi":"10.21203/rs.3.rs-7301811/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7301811/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eObjective\u003c/h2\u003e\u003cp\u003eTo establish a machine-learning model of left atrial appendage thrombogenic milieu (LAATM) in patients with nonvalvular atrial fibrillation (NVAF), and analyze the corresponding risk factors to guide clinical decision-making.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003ePatients with NVAF were selected and divided into LAATM group and non-LAATM group, according to the results of transesophageal echocardiography (TEE). The LAATM group included LAA thrombus formation, sludge and spontaneous echo contrast. The patient data was collected and preprocessed. The machine learning algorithms of random forest (RF), support vector machine (SVM) and extreme gradient boosting (XGBoost) were used to establish a predictive model for LAATM in patients with NVAF. Shapley additive explanation (SHAP) was used to sort the feature importance of clinical factors.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eA total of 1217 patients were selected in this study, including 112 patients in LAATM group and 1105 patients in non-LAATM group. In terms of predictive performance, AUC value of RF model was 0.97, F1 score was 0.93, accuracy was 0.98, precision was 0.99, and recall was 0.89; AUC value of SVM model is 0.96, F1 score is 0.89, accuracy is 0.97, precision is 0.95, recall is 0.84; The AUC value of the XGBoost model is 0.96, F1 score is 0.88, accuracy is 0.97, precision is 0.98, and recall is 0.82. The prediction efficiency of RF model is the best. The prediction results of RF model were visualized by SHAP diagram, indicating that Homocysteine (HCY), NT-proBNP, C-reactive protein(CRP), glycosylated hemoglobin (HbA1c) and ABC stroke score were the top five risk factors affecting the formation of LAATM in patients with NVAF.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e\u003cp\u003eThe RF model achieved the best predictive performance between the three prediction model. HCY, NT-proBNP, CRP, HbA1c and ABC stroke score were the top five risk factors affecting the formation of LAATM in patients with NVAF.\u003c/p\u003e","manuscriptTitle":"Research on left atrial appendage thrombogenic milieu prediction model in patients with nonvalvular atrial fibrillation based on machine learning algorithm","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-08 10:15:38","doi":"10.21203/rs.3.rs-7301811/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-09-15T05:04:07+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-13T10:05:33+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-05T08:28:49+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"217704503565773234752371987374884516083","date":"2025-08-30T04:10:42+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"39839529627609847043850688944943510902","date":"2025-08-29T08:15:03+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-08-28T10:30:08+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-08-28T06:31:48+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-08-19T07:11:51+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-08-18T15:46:39+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Cardiovascular Disorders","date":"2025-08-18T15:43:22+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-cardiovascular-disorders","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcar","sideBox":"Learn more about [BMC Cardiovascular Disorders](http://bmccardiovascdisord.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcar/default.aspx","title":"BMC Cardiovascular Disorders","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"673edfec-0c93-4ea2-8cc2-168d234a97a7","owner":[],"postedDate":"September 8th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-01-05T16:05:45+00:00","versionOfRecord":{"articleIdentity":"rs-7301811","link":"https://doi.org/10.1186/s12872-025-05396-y","journal":{"identity":"bmc-cardiovascular-disorders","isVorOnly":false,"title":"BMC Cardiovascular Disorders"},"publishedOn":"2025-12-30 15:58:07","publishedOnDateReadable":"December 30th, 2025"},"versionCreatedAt":"2025-09-08 10:15:38","video":"","vorDoi":"10.1186/s12872-025-05396-y","vorDoiUrl":"https://doi.org/10.1186/s12872-025-05396-y","workflowStages":[]},"version":"v1","identity":"rs-7301811","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7301811","identity":"rs-7301811","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}