|
徐志云.急性心肌梗死急诊经皮冠状动脉介入术后预后不良危险因素分析：一项随机森林模型研究[J].内科急危重症杂志,2025,31(6):545-548
|
| DOI：10.11768/nkjwzzzz20250612 |
| 中文关键词: 急性心肌梗死 急诊经皮冠状动脉介入治疗 预后 风险分析 随机森林模型 |
| 英文关键词: |
| 基金项目:南通市卫生健康委员会科研课题（MSZ2022031） |
|
| 摘要点击次数: 6 |
| 全文下载次数: 7 |
| 中文摘要: |
| 摘要 目的：基于随机森林模型分析急性心肌梗死（AMI）患者行急诊经皮冠状动脉介入（PCI）术后的预后风险。方法：连续选取急诊经PCI治疗的275例AMI患者，以预后不良为终点事件，最终纳入270例患者，将其分为预后不良组42例和预后良好组228例，收集患者入院时一般资料、相关检验指标、围术期指标。将可能影响AMI患者急诊PCI术后预后不良的变量纳入随机森林模型，利用精确系数、基尼系数综合评估变量的重要程度，十折交叉验证模型性能。结果：术后随访至2023年10月，中位随访时间14.5个月（7~28个月）。共270例获得随访，预后不良发生率约为15.56%（42/270），其中心源性死亡7例（16.67%）、再发AMI 12例（28.57%）、频发室颤再入院15例（35.71%）、急性心功能不全再入院8例（19.05%）。与预后良好组比较，预后不良组年龄≥60岁患者比例、高血压史患者比例及甘油三酯（TG）、中性粒细胞明胶酶相关脂质运载蛋白（NGAL）、同型半胱氨酸（Hcy）、血肌酐（Scr）、肌酸激酶同工酶（CK-MB）、白细胞计数（WBC）、血小板平均容积（MPV）水平更高，左心室射血分数（LVEF）更低（P均＜0.05）。基于mtry=4、ntree=400构建随机森林模型，综合各变量对模型预测的精确度（＞10）和基尼系数（＞100）显示，NGAL、MPV、年龄、Hcy、高血压史是影响AMI患者急诊PCI术后预后不良的前五位重要变量。十折交叉验证对最终建立的随机森林模型的内部验证结果显示，该模型准确率、曲线下面积（AUC）分别为0.852、0.903，敏感度、特异性分别为87.20%、81.00%，各参数值均＞0.80，提示模型预测真实可靠。结论：NGAL、MPV、年龄、Hcy、高血压史是影响AMI患者急诊PCI术后预后不良的危险因素，可为临床评估患者术后获益与风险提供参考依据。 |
| 英文摘要: |
| Abstract Objective: To analyze the prognostic risk of patients with acute myocardial infarction (AMI) following emergency percutaneous coronary intervention (PCI) based on a random forest model. Methods: A total of 275 AMI patients who underwent emergency PCI were consecutively selected, with poor prognosis as the endpoint event. Ultimately, 270 patients were included and divided into a poor prognosis group (42 cases) and a good prognosis group (228 cases). General data, relevant laboratory indicators, and perioperative indicators were collected upon admission. Variables potentially influencing poor prognosis after emergency PCI in AMI patients were incorporated into the random forest model. The importance of variables was comprehensively assessed using accuracy and Gini coefficients, and model performance was validated through ten-fold cross-validation. Results: Follow-up was conducted until October 2023, with a median follow-up duration of 14.5 months (range: 7-28 months). Among the 270 patients followed up, the incidence of poor prognosis was approximately 15.56% (42/270), including 7 cases of cardiac death (16.67%), 12 cases of recurrent AMI (28.57%), 15 cases of frequent ventricular fibrillation requiring rehospitalization (35.71%), and 8 cases of acute cardiac insufficiency requiring rehospitalization (19.05%). Compared with the group with good prognosis,the proportion of patients with age≥60 years old, the proportion of patients with history of hypertension, the levels of triglycerides (TG), neutrophil gelatinase-associated lipocalin (NGAL), homocysteine (Hcy), serum creatinine (Scr), creatine kinase-MB (CK-MB), white blood cell count (WBC), and mean platelet volume (MPV) in the poor prognosis group were increased, and left ventricular ejection fraction (LVEF) was relatively decreased (all P< 0.05). Based on mtry=4 and ntree=400, the random forest model was constructed. The accuracy (>10) and Gini coefficient (>100) of the variables indicated that NGAL, MPV, age, Hcy, and history of hypertension were the top five important variables influencing poor prognosis after emergency PCI in AMI patients. The internal validation results of the final random forest model through ten-fold cross-validation showed that the model's accuracy and area under the curve (AUC) were 0.852 and 0.903, respectively, with sensitivity and specificity of 87.20% and 81.00%, respectively. All parameter values were >0.80, indicating the model's reliable predictive performance. Conclusion: NGAL, MPV, age, Hcy, and history of hypertension are risk factors influencing poor prognosis after emergency PCI in AMI patients, providing a reference for clinical assessment of postoperative benefits and risks. |
|
|
|
|
