| 心房颤动合并冠心病患者抗栓强度预测模型的构建与验证 |
| 投稿时间：2025-07-28 修订日期：2025-08-17 点此下载全文 |
| 引用本文：杜荣生,闫增强,路林峰,孙玉翠,张彪.心房颤动合并冠心病患者抗栓强度预测模型的构建与验证[J].医学研究杂志,2025,54(12):140-146 |
| DOI：
10.11969/j.issn.1673-548X.2025.12.024 |
| 摘要点击次数: 21 |
| 全文下载次数: 14 |
|
| 基金项目:河北省沧州市重点研发计划项目(222106137) |
|
| 中文摘要:目的 构建并验证基于临床特征和出血风险分层的心房颤动合并冠心病患者抗栓强度预测模型,评估其指导个体化抗栓治疗的临床价值。方法 采用多中心、前瞻性队列研究方法,选取2021年1月～2023年6月沧州市人民医院心内科收治的1847例心房颤动合并冠心病患者为研究对象,根据实际抗栓强度分为单抗组(n=412)、双抗组(n=968)和三联组(n=467)。基于改良心房颤动抗凝治疗出血风险评分系统(hypertension, abnormal renal/liver function, stroke, bleeding history or predisposition, labile INR, elderly (>65), drugs/alcohol concomitantly,HAS-BLED)评分联合血小板功能检测和基因多态性分析进行出血风险分层,筛选预测因子,构建抗栓强度预测模型。将2023年7月～2024年6月沧州市人民医院心内科收治的428例患者应用随机数字表分为模型指导治疗组(n=216)和常规治疗组(n=212),随访12个月评估其临床结局。结果 出血风险分层显示,低危587例、中危892例、高危368例,筛选出18个独立预测因子。构建的模型整体预测准确率为82.6%,单抗、双抗、三联治疗的预测曲线下面积(area under the curve,AUC)值分别为0.892、0.856、0.901。经皮冠状动脉介入治疗Taxus支架与心脏外科手术的协同作用(synergy between PCI with taxus and cardiac surgery,SYNTAX)评分、HAS-BLED评分、全球急性冠状动脉事件注册(global registry of acute coronary events, GRACE)评分、血小板聚集率和CYP2C19基因型是影响抗栓强度选择的五大核心因素。验证发现,模型指导治疗组主要不良心脑血管事件发生率显著低于常规治疗组(8.3% vs 14.2%,P=0.022),大出血事件发生率亦明显降低(4.2% vs 8.5%,P=0.037)。结论 基于临床特征和出血风险分层的抗栓强度预测模型能够准确预测心房颤动合并冠心病患者的最适抗栓治疗强度,为精准抗栓治疗提供了有效工具。 |
| 中文关键词:心房颤动 冠心病 抗栓治疗 机器学习 个体化医疗 |
| |
| Construction and Validation of Antithrombotic Intensity Prediction Model for Patients with Atrial Fibrillation and Coronary Artery Disease. |
|
|
| Abstract:Objective ObjectiveTo construct and validate an antithrombotic intensity prediction model for patients with atrial fibrillation (AF) and coronary artery disease (CAD) based on clinical features and bleeding risk stratification, and to evaluate its clinical value in guiding individualized antithrombotic therapy. Methods A multicenter, prospective cohort study was conducted. A total of 1847 patients with AF and CAD admitted to Cangzhou People′s Hospital Department of Cardiology from January 2021 to June 2023 were enrolled, and there were divided into monotherapy (n=412), dual therapy (n=968), and triple therapy (n=467) groups according to actual antithrombotic intensity. Bleeding risk stratification was performed using modified hypertension,abnormal renal and liver function,stroke,bleeding,labile international normalized ratio,elderly,drugs and alcohol intake (HAS-BLED) score combined with platelet function testing and genetic polymorphism analysis. Predictive factors were screened to construct an antithrombotic intensity prediction model. A total of 428 patients admitted to Cangzhou People′s Hospital Department of Cardiology from July 2023 to June 2024 were randomly into model-guided therapy group (n=216) and conventional therapy group (n=212) with randoml number table, with 12-month follow-up to assess clinical outcomes. Results Bleeding risk stratification identified 587 low-risk, 892 intermediate-risk, and 368high-risk patients. Eighteen independent predictors were identified. The model achieved an overall prediction accuracy of 82.6%, with area under the curve (AUC) values of 0.892,0.856 and 0.901 for monotherapy, dual therapy, and triple therapy predictions, respectively. SYNergy between PCI with TAXus and cardiac surgery (SYNTAX) score, HAS-BLED score, Global Registry of Acute Coronary Events (GRACE) score, platelet aggregation rate, and CYP2C19genotype were the five core factors influencing antithrombotic intensity selection. In validation, verification revealed thatthe model-guided therapy group had significantly lower incidence of major adverse cardiovascular and cerebrovascular events compared to the conventional therapy group (8.3% vs 14.2%, P=0.022), and the incidence of major bleeding events was also significantly reduced(4.2% vs 8.5%, P=0.037). Conclusion The antithrombotic intensity prediction model based on clinical features and bleeding risk stratification can accurately predict optimal antithrombotic therapy intensity for patients with AF and CAD, providing an effective tool for precision antithrombotic treatment. |
| keywords:Atrial fibrillation Coronary artery disease Antithrombotic therapy Machine learning Individualized medicine |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
