目的 基于机器学习方法构建自体动静脉内瘘血栓形成风险预测模型并进行模型验证。 方法 以2020年3月—2021年12月在哈尔滨医科大学附属第二医院血液净化中心行维持性血液透析的患者为研究对象，应用逻辑回归(logistic regression，Logistic)和随机森林(random forest，RF)构建模型。绘制各模型受试者工作特征曲线(receiver operating characteristic curve，ROC)和受试者工作特征曲线下面积(area under the curve，AUC),并用准确率、特异度、灵敏度和F1度量评价模型性能。 结果 270例MHD患者中，AVF血栓形成组105例(38.89%)，非AVF血栓形成组165例(61.11%)，最终纳入吸烟史(OR=2.992，95% CI：1.306～6.854，P=0.010)、高血压史(OR=12.376，95% CI：3.432～44.624，P＜0.001)、糖尿病史(OR=7.477，95% CI：2.887～19.360，P＜0.001)、高血脂史(OR=6.947，95% CI：2.733～17.659，P＜0.001)、冠心病史(OR=12.894，95% CI：4.827～34.439，P＜0.001)、穿刺点压迫时间(OR=1.132，95% CI：1.053～1.217，P=0.010)、三酰甘油(OR=1.322，95% CI：1.005～1.741，P=0.046)等7个因素构建风险预测模型。RF预测模型的AUC为0.944，Logistic模型的AUC为0.895(Z=1.688，P=0.092)。 结论 吸烟史、高血压史、糖尿病史、高血脂史、冠心病史、穿刺点压迫时间和三酰甘油是MHD患者发生AVF血栓形成的高危因素，基于上述危险因素构建的2种预测模型性能良好，可互相补充。
Objective To construct a risk prediction model for thrombosis in autogenous arteriovenous fistula (AVF) based on machine learning, and to verify the model. Methods A total of 270 patients undergoing maintenance hemodialysis (MHD) in Hemodialysis Center, The Second Affiliated Hospital of Harbin Medical University from March 2020 to December 2021 were recruited as the study subjects. Logistic regression and random forest were used to construct the models. Receiver operating characteristic curve and area under the curve (AUC) were plotted for each model. AUC, accuracy, specificity, sensitivity and F1-score were used to evaluate the models. Results Among 270 MHD patients, 105 cases (38.89%) were in AVF thrombosis group and 165 cases (61.11%) in non-AVF thrombosis group. The seven risk factors of smoking history (OR=2.992, 95% CI: 1.306～6.854, P=0.010), hypertension history (OR=12.376, 95% CI: 3.432～44.624, P<0.001), diabetes history (OR=7.477, 95% CI: 2.887～19.360, P<0.001), hyperlipidemia history (OR=6.947, 95% CI: 2.733～17.659, P<0.001), coronary heart disease history (OR=12.894, 95% CI: 4.827～34.439, P<0.001), puncture point compression time (OR=1.132, 95% CI: 1.053～1.217, P=0.010), and triglyceride (OR=1.322, 95% CI: 1.005～1.741, P=0.046) were used to construct risk prediction models. The area under the curve of random forest prediction model was 0.944, and that of logistic regression model was 0.895 (Z=1.688, P=0.092). Conclusion Smoking history, hypertension history, diabetes history, hyperlipidemia history, coronary heart disease history, puncture point compression time and triacylglycerol are high risk factors for thrombosis in AVF in MHD patients. The two models based on the seven risk factors have good predictive performance and can be complementary each other.
[1]L-A Inker, Astor B-C, Fox C-H, et al. KDOOI US commentary on the 2012 KDIGO clinical practice guideline for the evaluation and management of CKD[J]. Am J Kidney Dis, 2014, 63(5)：713-735.
[2]Wang L, Xu X, Zhang M, et al. Prevalence of Chronic Kidney Disease in China: Results From the Sixth China Chronic Disease and Risk Factor Surveillance. JAMA Intern Med. Published online February 20, 2023.
[3]陈晓磊, 冯杰, 秦怡.慢性尿毒症患者血液透析治疗方式的选择及有效性对照[J]. 中国医药指南, 2020,18(02):17-18.
[4]左力.血液净化模式选择专家共识[J]. 中国血液净化, 2019, 18(07): 442-472.
[5]Fang W, Chao Y, Jianyan L, et al. Executive summary for the 2015 Annual Data Report of? the China Kidney Disease Network (CK-NET)[J]. Kidney International, 2019,95(3):501-505.
[6]Shiu YT，Rotmans JI，Geelhoed WJ. Arteriovenous conduits for hemodialysis：how to better modulate the pathophysiological vascular response to optimize vascular access durability. Am J Physiol Renal Physiol，2019，316（5）：F794-F806.
[7]Lee T. Fistula first initiative：historical impact on vascular access practice patterns and influence on future vascular access care. Cardiovasc Eng Technol，2017，8（3）：244-254.
[8] Quencer KB, Friedman T. Declotting the thrombosed ac‐ cess[J]. Tech Vasc Interv Radiol, 2017, 20(1):38-47.
[9]王玉柱,叶朝阳,金其庄.中国血液透析用血管通路专家共识(第1版).中国血液净化,2014,13(08):549-558.
[10]HS，Yeong-Chang C，Ming-Hsing H. Application of bioimpedance spectroscopy in Asian Dialysis patients（ABISAD-III）：a randomized controlled trial for clinical outcomes.Int Urol Nephrol，2016，48（11）：1897-1909.
[11]Takashima T，Matsumoto K，Takeshita C. Restoration of autologous arteriovenous fistula by removal of the occluded short venous part and venovenous end-to-end anastomosis in a hemodialysis patient.Ann Vasc Surg，2018，52：313.
[12]孔祥静,肖达平,徐静琳等.尿毒症血液透析患者血浆ADAMTS-13变化与动静脉内瘘血栓形成的关系[J].浙江中西医结合杂志,2021,31(07):621-625.
[13]钟小燕,杨枫,钟晓容等.CD62P、CD63、平均血小板体积与动静脉内瘘血栓形成的关系研究[J].中外医疗,2021,40(29):31-34.
[14]李莉,薛莲,伍梦寒.糖尿病肾病患者凝血功能血浆血红蛋白和D-D水平及其与血液透析动静脉内瘘血栓形成的关系研究[J].河北医学,2023,29(01):107-112.
[15]杨玲，刘斌，于俊超.维持性血液透析患者自体动静脉内瘘血栓影响因素观察.临床肾脏病杂志，2016，16 (2) :96-99.
[16]马鹏跃,吴永东,黄凤璋等.血液透析患者动静脉内瘘晚期失功影响因素及其预测效能[J].山东医药,2021,61(12):81-84.
[17] Cheung AK, Imrey PB, Alpers CE, et al. Intimal Hyper‐ plasia, Stenosis, and Arteriovenous Fistula Maturation Failure in the Hemodialysis Fistula Maturation Study [J]. J Am Soc Nephrol, 2017,28(10):3005-3013.
[18] Lazarides MK, Georgiadis GS, Antoniou GA, et al. A metaanalysis of dialysis access outcome in elderly patients [J]. J Vasc Surg, 2007,45(2):420-426.
[19]张宇涵,马岩,彭宇等.自体动静脉内瘘血栓危险因素的Meta分析[J].中国血液净化,2021,20(02):126-129.
[20]Staaf K，Uhlin F. Cannulation with sharp or blunt needles for haemodialysis：the importance of cannulation technique for the patient's lifeline. J Vasc Access，2019，20（2）：161-168.
[21]Qin HY，Jia P，Liu H. Nursing strategies for patients with chronic renal failure undergoing maintenance hemodialysis treatment by arteriovenous fistula.Iran J Public Health，2016，45（10）：1270-1275.
[22]张一页,鄢艳.血液透析患者自体动静脉内瘘临床预测模型研究现状概述[J].中国血液净化,2023,22(03):198-201.
[23]郑诗,梅游英,王诣涵等.随机森林模型和logistic回归模型在维持性血液透析患者动静脉内瘘失功预测中的应用效果比较[J].中国医院统计,2021,28(06):485-490.
[24] Peralta R, Garbelli M, Bellocchio F, et al. Development and validation of a machine learning model predicting arteriovenous fistula failure in a large network of dialysis clinics[J]. Int J Environ Res Public Health, 2021,18(23):12355.
