A risk model based on miR-483-5p and miR-150 for atrial fibrillation recurrence

Figures

• 

  Fig. 1. Expression levels and expression differences of miR-483-5p and miR-150 in AF. A Expression levels of miR-483-5p in circulating samples of the control group and patients with AF (All data were obtained from databases). B Expression levels of miR-150 in circulating samples of the control group and patients with AF. C Log2 fold change in miRNA expression of patients with AF relative to healthy controls. (n represents the number of independent samples in each GEO dataset; see Methods for details)

• 

  Fig. 2. Comparison of clinical characteristics and biomarkers between AF and control groups. A Comparison of case distribution between the study group (orange bars) and the control group (green bars) in the subgroups of paroxysmal atrial fibrillation and persistent atrial fibrillation. B Intergroup comparison of the relative expression levels of plasma miR-483-5p (green bars), plasma miR-150 (blue bars), and LVEF (yellow bars) between the study group and the control group; the left vertical axis represents the relative expression of miRNAs, the right vertical axis represents LVEF percentage, and the error bars represent mean ± standard deviation (n = 122 for AF group; n = 60 for control group; independent patient samples). LVEF: Left ventricular ejection fraction

• 

  Fig. 3. Correlation between miRNA expression and LVEF in AF patients. A Bar plot of Pearson correlation coefficients. Plasma miR-483-5p expression was negatively correlated with LVEF (r = −0.383, P < 0.001), while miR-150 expression was positively correlated with LVEF (r = 0.479, P < 0.001). LVEF: Left Ventricular Ejection Fraction. B Scatter plot showing the negative linear association between plasma miR-483-5p and LVEF. C Scatter plot showing the positive linear association between plasma miR-150 and LVEF (n = 122 patients; each dot represents one independent patient sample)

• 

  Fig. 4. Distribution of comorbidities in recurrence and non-recurrence groups. A Proportion of comorbidities in the non-recurrence group (DM: Diabetes Mellitus, HBP: High Blood Pressure, CAHD: Coronary Atherosclerotic Heart Disease). B Proportion of comorbidities in the recurrence group (n = 36 recurrence patients; n = 86 non-recurrence patients)

• 

  Fig. 5. Plasma miR-483-5p, miR-150, and their combined detection for disease diagnosis-related analysis. A ROC curves of plasma miR-483-5p, plasma miR-150, and their combined detection. The purple line is the random diagnosis reference line (AUC = 0.5). B Bar graph showing the AUC of the three detection modes (plasma miR-483-5p, plasma miR-150, and their combination). AUC: Area Under the Curve. C Bar graph comparing sensitivity (orange columns) and specificity (green columns) of the three detection methods. D Bar graph presenting the Youden index of plasma miR-483-5p, plasma miR-150, and their combined detection (n = 122 patients; independent samples used for ROC analysis)

• 

  Fig. 6. Risk stratification model based on miR-483-5p and miR-150 expression. A The dashed lines represent the optimal classification thresholds (miR-483-5p: 1.025; miR-150: 0.805). B Box plots of miRNA expression levels in different risk groups, left panel: miR-483-5p expression levels, right panel: miR-150 expression levels. C The y-axis represents the recurrence rate (%), and the x-axis represents three risk grades. The specific recurrence rate values are labeled on the top of the bars, and the sample size of each group is labeled inside the bars. ***P < 0.001, ****P < 0.0001. D Confusion matrix of the risk stratification model, the rows represent the predicted risk grades (high, medium, low), and the columns represent the actual outcomes (recurrence, non-recurrence). The darker the color, the greater the number of patients in the corresponding category. E Key performance metrics of the risk stratification model, the y-axis represents the values of each metric, and the x-axis corresponds to the four metrics (Accuracy, Precision, Recall, F1 Score), n = 38 high-risk, n = 42 medium-risk, n = 42 low-risk; independent patient samples