基于决策树算法构建急性心肌梗死患者PCI术后血运重建的风险预测方案

doi:10.3760/cma.j.cn441417-20240621-03004

摘要/Abstract

摘要：

目的　运用决策树算法构建急性心肌梗死（AMI）患者经皮冠状动脉介入治疗（PCI）术后血运重建的风险预测模型。方法　回顾性分析2021年1月至2023年1月在陕西中医药大学第二附属医院行PCI术的203例AMI患者临床资料，根据术后1年内有无再次血运重建分为血运重建组（60例）和非血运重建组（143例）。血运重建组男41例，女19例，年龄（62.75±10.32）岁。非血运重建组男94例，女49例，年龄（61.47±10.07）岁。采用多因素logistic回归分析法探讨AMI患者PCI术后再次血运重建的影响因素。按照7∶3比例将203例患者随机分为训练集（142例）和测试集（61例），基于训练集数据构建决策树模型，基于测试集数据验证决策树模型的预测效能。采用χ²检验、t检验进行统计分析。结果　血运重建组的糖尿病、PCI术前低密度脂蛋白胆固醇（LDL-C）≥3.4 mmol/L、尿酸>420 µmol/L、超敏C反应蛋白（hs-CRP）>10 mg/L及病变支数≥2支、支架数量≥3个、PCI术后残余SYNTAX评分（rSS）>5分患者占比均高于非血运重建组［23.33%（14/60）比11.19%（16/143）、36.67%（22/60）比20.98%（30/143）、38.33%（23/60）比20.98%（30/143）、33.33%（20/60）比17.48%（25/143）、70.00%（42/60）比53.85%（77/143）、61.67%（37/60）比45.45%（65/143）、38.33%（23/60）比21.68%（31/143）］，差异均有统计学意义（均P<0.05）。多因素logistic回归分析结果显示，糖尿病、PCI术前LDL-C≥3.4 mmol/L、尿酸>420 µmol/L、hs-CRP >10 mg/L及病变支数≥2支、PCI术后rSS>5分均为AMI患者PCI术后再次血运重建的危险因素（均P<0.05）。基于142例训练集数据建立AMI患者PCI术后再次血运重建的决策树风险预测模型，筛选出PCI术前尿酸、hs-CRP、LDL-C水平及PCI术后rSS、糖尿病、病变支数6个解释变量，提取7条分类规则，其中PCI术前尿酸水平为该模型的首要影响因素。基于61例测试集数据对决策树模型进行验证，结果显示该模型预测AMI患者PCI术后再次血运重建的灵敏度为88.89%，特异度为83.72%，准确度为85.25%。结论　AMI患者PCI术后再次血运重建的决策树风险模型包含6个变量，分别为PCI术前尿酸、hs-CRP、LDL-C水平及PCI术后rSS、糖尿病、病变支数，PCI术前尿酸水平为该模型的首要影响因素。

关键词:

急性心肌梗死, 经皮冠状动脉介入治疗, 血运重建, 影响因素, 决策树模型

Abstract:

Objective　To construct a risk prediction model for revascularization in patients with acute myocardial infarction (AMI) after percutaneous coronary intervention (PCI) using decision tree algorithm. Methods　The clinical data of 203 patients with AMI who underwent PCI in the Second Affiliated Hospital of Shaanxi University of Traditional Chinese Medicine from January 2021 to January 2023 were retrospectively analyzed, and they were divided into a revascularization group (60 cases) and a non-revascularization group (143 cases) according to whether they underwent revascularization within 1 year after operation. In the revascularization group, there were 41 males and 19 females, aged (62.75±10.32) years. In the non-revascularization group, there were 94 males and 49 females, aged (61.47±10.07) years. Multivariate logistic regression analysis was used to investigate the influencing factors of revascularization in AMI patients after PCI. According to the ratio of 7:3, 203 patients were randomly divided into a training set (142 cases) and a test set (61 cases). The decision tree model was constructed based on the training set data, and the prediction efficiency of the decision tree model was verified based on the test set data. χ² test and t test were used for statistical analysis. Results　In the revascularization group, the proportions of diabetes mellitus, low density lipoprotein cholesterol (LDL-C) ≥3.4 mmol/L, uric acid >420 µmol/L, hypersensitive C-reactive protein (hs-CRP) >10 mg/L, number of lesions ≥2, and number of stents ≥3 before PCI, and residual SYNTAX score (rSS) >5 points after PCI were higher than those in the non-revascularization group [23.33% (14/60) vs. 11.19% (16/143), 36.67% (22/60) vs. 20.98% (30/143), 38.33% (23/60) vs. 20.98% (30/143), 33.33% (20/60) vs. 17.48% (25/143), 70.00% (42/60) vs. 53.85% (77/143), 61.67% (37/60) vs. 45.45% (65/143), 38.33% (23/60) vs. 21.68% (31/143)], with statistically significant differences (all P<0.05). Multivariate logistic regression analysis showed that diabetes mellitus, LDL-C ≥3.4 mmol/L, uric acid >420 µmol/L, hs-CRP >10 mg/L, and number of lesions ≥2 before PCI, and rSS >5 points after PCI were all risk factors for revascularization in AMI patients after PCI (all P<0.05). Based on the training set data, a decision tree risk prediction model for revascularization in AMI patients after PCI was established. The model contained 6 explanatory variables, namely diabetes mellitus, LDL-C level, uric acid level, hs-CRP level, and number of lesions before PCI, and rSS after PCI. A total of 7 classification rules were extracted, among which the uric acid level before PCI was the primary influencing factor of the model. The decision tree model was verified based on the test set data, and the sensitivity, specificity, and accuracy of the decision tree model for predicting revascularization in AMI patients after PCI were 88.89%, 83.72%, and 85.25%, respectively. Conclusion　The decision tree risk model of revascularization in AMI patients after PCI includes 6 variables, namely diabetes mellitus, LDL-C level, uric acid level, hs-CRP level, and number of lesions before PCI, and rSS after PCI, among which the uric acid level before PCI is the primary influencing factor of the model.

Key words:

Acute myocardial infarction, Percutaneous coronary intervention, Revascularization, Influencing factors, Decision tree model

翟夏康启赵学飞李敏杰陈敏娜董欢乐董静.

基于决策树算法构建急性心肌梗死患者PCI术后血运重建的风险预测方案 [J]. 国际医药卫生导报, 2025, 31(3): 370-376.

Zhai Xia, Kang Qi, Zhao Xuefei, Li Minjie, Chen Minna, Dong Huanle, Dong Jing.

Constructing a risk prediction plan for revascularization in patients with acute myocardial infarction after PCI based on decision tree algorithm [J]. International Medicine and Health Guidance News, 2025, 31(3): 370-376.

参考文献

[1] Lee SH, Jeong MH, Ahn JH, et al. Predictors of recurrent acute myocardial infarction despite successful percutaneous coronary intervention[J]. Korean J Intern Med, 2022, 37(4):777-785. DOI: 10.3904/kjim.2021.427.
[2] 李丽. AMI患者PCI术后血清BNP、hs-CRP变化及其与预后的关系[J]. 国际医药卫生导报,2022,28(12):1715-1719. DOI:10.3760/cma.j.issn.1007-1245.2022.12.020.
[3] Liu F, Liu XJ, He YP, et al. Clinical value of GRACE score combined with DFR in predicting short-term prognosis of patients undergoing early PCI after thrombolysis for acute myocardial infarction[J]. Eur Rev Med Pharmacol Sci, 2023, 27(9): 4038-4045. DOI: 10.26355/eurrev_202305_32309.
[4] 范鹏云 .24 h动态心电图参数评估急性心肌梗死患者PCI治疗后恶性心律失常的研究[J].国际医药卫生导报,2022,28(6):863-866.DOI:10.3760/cma.j.issn.1007-1245.2022. 06.032.
[5] 尉驰,闫浩东,徐力红,等. ACEF评分对急性冠状动脉综合征患者经皮冠状动脉介入治疗术后6个月内再次血运重建的评估价值[J]. 中国心血管杂志,2019,24(6):501-506. DOI:10.3969/j.issn.1007-5410.2019.06.004.
[6] 周晓丹,王岩,项军,等. 肌钙蛋白、SNYTAX积分、GRACE评分与老年冠心病PCI术后再次血运重建的关系[J]. 疑难病杂志,2023,22(8):824-828. DOI:10.3969/j.issn.1671- 6450.2023.08.007.
[7] 任晓园,王慧峰,张强,等 .一氧化氮合酶1转接蛋白基因多态性与冠心病病人PCI术后需要再次血运重建的关系[J].中西医结合心脑血管病杂志,2021,19(17):2982-2986.DOI:10.12102/j.issn.1672-1349.2021.17.023.
[8] Xiang S, Li L, Wang L, et al. A decision tree model of cerebral palsy based on risk factors[J]. J Matern Fetal Neonatal Med, 2021, 34(23):3922-3927. DOI: 10.1080/14767058.2019.1702944.
[9] 中华医学会心血管病学分会,中华心血管病杂志编辑委员会,«中国循环杂志»编辑委员会 .急性心肌梗死诊断和治疗指南[J].中华心血管病杂志,2001,29(12):710-725.DOI:10.3760/j:issn:0253-3758.2001.12.003.
[10] M A, Khandait H, Guralwar C, et al. A Study on QT dispersion before and after thrombolysis in acute myocardial infarction and its prognostic implications: a before and after comparison study[J]. J Assoc Physicians India, 2023, 71(10):14-18. DOI: 10.59556/japi.71.0363.
[11] Sawayama Y, Takashima N, Harada A, et al. Incidence and in-hospital mortality of acute myocardial infarction: a report from a population-based registry in Japan[J]. J Atheroscler Thromb, 2023, 30(10):1407-1419. DOI: 10.5551/jat.63888.
[12] Saito Y, Oyama K, Tsujita K, et al. Treatment strategies of acute myocardial infarction: updates on revascularization, pharmacological therapy, and beyond[J]. J Cardiol, 2023, 81(2):168-178. DOI: 10.1016/j.jjcc.2022.07.003.
[13] 李涵,黄宇鹏,金红艳,等 .外周血APN、HbA1c、糖基化ApoA-I水平与冠心病患者PCI术后血运重建的关系[J].解放军医药杂志,2021,33(9):58-61,68.DOI:10.3969/j.issn.2095-140X.2021.09.012.
[14] 曹仲颖,刘培培,王景祥,等. PCI术后再次血运重建的危险因素分析[J]. 中西医结合心脑血管病杂志,2021,19(15):2619-2622. DOI:10.12102/j.issn.1672-1349.2021.15.027.
[15] Kang L, Xie D, Chen D, et al. Prognostic value of NT-proBNP and uric acid in acute ST-segment elevation myocardial infarction patients after complete revascularization[J]. Am J Transl Res, 2024, 16(8):4182-4189. DOI: 10.62347/VQWS9174.
[16] Kajana V, Massaro G, Somaschini A, et al. Serum uric acid in patients with ST-segment elevation myocardial infarction: an innocent bystander or leading actor?[J]. Nutr Metab Cardiovasc Dis, 2022, 32(7):1583-1589. DOI: 10.1016/j.numecd.2022.03.023.
[17] Xia JG, Li B, Zhang H, et al. Precise metabolomics defines systemic metabolic dysregulation distinct to acute myocardial infarction associated with diabetes[J]. Arterioscler Thromb Vasc Biol, 2023, 43(4):581-596. DOI: 10.1161/ATVBAHA.122.318871.
[18] 张琳赟,刘杰,吴皖灵,等. 列线图模型预测冠心病患者经皮冠状动脉介入治疗术后因严重支架内再狭窄导致再次血运重建的研究[J]. 中国介入心脏病学杂志,2021,29(5):281-287. DOI:10.3969/j.issn.1004-8812.2021.05.007.
[19] Choi WG, Baek MJ, Rha SW, et al. Impact of initial very low-level low-density lipoprotein cholesterol on the prognosis of acute myocardial infarction patients[J]. Coron Artery Dis, 2022, 31(1):e44-e50. DOI: 10.1097/MCA.0000000000001085.
[20] 中华人民共和国国家卫生健康委员会.成人高脂血症食养指南(2023年版)[J].全科医学临床与教育,2023,21(7):581-583.DOI:10.13558/j.cnki.issn1672-3686.2023. 007.002.
[21] 廖贵红,韦波,韩运峰. 急性心肌梗死患者血清表皮调节素水平与患者预后关系研究[J]. 中国循证心血管医学杂志,2020,12(12):1496-1499. DOI:10.3969/j.issn.1674-4055. 2020.12.20.
[22] Miao J, Du T. The relationship between levels of tumor necrosis factor-alpha, interleukin-6, and C-reactive protein in the serum of elderly and acute myocardial infarction[J]. Cell Mol Biol (Noisy-le-grand), 2024, 70(4):95-99. DOI: 10.14715/cmb/2024.70.4.15.
[23] 陈亚栋,梁燕芳,冼朝丽 .急性心肌梗死经PCI治疗后D-二聚体、hs-CRP和LDL-C水平与预后的相关性[J].海南医学,2022,33(4):421-423.DOI:10.3969/j.issn.1003-6350. 2022.04.004.
[24] 刘晓红,李为民 .残余SYNTAX评分对急性心肌梗死患者经皮冠状动脉介入治疗术后不良预后的预测价值[J].中国介入心脏病学杂志,2019,27(8):440-446.DOI:10.3969/j.issn.1004-8812.2019.08.004.
[25] Wang K, Liu Q, Mo S, et al. A decision tree model to help treatment decision-making for severe spontaneous intracerebral hemorrhage[J]. Int J Surg, 2024, 110(2):788-798. DOI: 10.1097/JS9.0000000000000852.
[26] Flayer CH, Perner C, Sokol CL. A decision tree model for neuroimmune guidance of allergic immunity[J]. Immunol Cell Biol, 2021, 99(9):936-948. DOI: 10.1111/imcb.12486.