您的位置:山东大学 -> 科技期刊社 -> 《山东大学学报(医学版)》

山东大学学报 (医学版) ›› 2020, Vol. 58 ›› Issue (6): 28-33.doi: 10.6040/j.issn.1671-7554.0.2019.1442

• • 上一篇    

高斯过程模型对慢性心衰患者1年内再入院的风险评估

张传备1*,李方2*,翟春晓3,余永明4,舒明雷5,王艺丹3,徐良栋6,郝恩魁3   

  • 发布日期:2022-09-27
  • 通讯作者: 郝恩魁. E-mail:haoenkui@sdu.edu.cn*共同第一作者
  • 基金资助:
    2014年山东省科技发展计划(2014GSF118187)

Gaussian process model for risk assessment of readmission for patients with chronic heart failure within one year

ZHANG Chuanbei1*, LI Fang2*, ZHAI Chunxiao3, YU Yongming4, SHU Minglei5, WANG Yidan3, XU Liangdong6, HAO Enkui3   

  1. 1. Clinical Medical College of Weifang Medical College, Weifang 261000, Shandong, China;
    2. Department of Health, Jinan Central Hospital, Jinan 250012, Shandong, China;
    3. Department of Cardiology, Shandong Provincial Qianfoshan Hospital, Jinan 250012, Shandong, China;
    4. School of Control Science and Materials, Shandong University, Jinan 250012, Shandong, China;
    5. National Supercomputer Center in Jinan, Jinan 250012, Shandong, China;
    6. Emergency Department, Dezhou Peoples Hospital, Dezhou 253000, Shandong, China
  • Published:2022-09-27

摘要: 目的 探讨通过使用高斯过程模型(GPM)和超声心动图参数建立的风险预测模型对慢性左室收缩功能减低(LVSD)心衰患者进行1年再入院风险评估的临床价值。 方法 收集并整理慢性LVSD患者290例,以1年内再入院为本研究终点。所有患者行常规超声心动图检查,并收集二尖瓣反流、三尖瓣反流、胸腔积液、心包积液、肺动脉收缩压、左心室内径及左室射血分数等参数数据。将290例患者随机分为70%训练数据和30%测试数据,使用机器学习算法对训练数据集进行信息交互分析,评估各项参数的重要性,并融合多超声参数建立GPM预测系统模型。随后利用GPM风险预测系统模型对测试数据集患者1年内再入院风险进行分析,将所有患者纳入超声心动图参数积分系统(超声积分系统)进行预测,最后通过受试者工作特征曲线(ROC)对两种方法进行分析比较。 结果 利用GPM信息交互分析各项超声心动图参数的权重分别为:二尖瓣反流23.64%,三尖瓣反流22.09%,胸腔积液16.18%,心包积液14.36%,肺动脉收缩压9.04%,左心室内径8.86%,左室射血分数5.83%。基于GPM的风险预测系统与超声积分系统的受试者工作特征曲线下面积(AUC)分别为83.10%(95%CI: 0.797~0.864)和70.60%(95%CI: 0.647~0.765)。 结论 基于超声心动图参数的高斯过程模型能够很好的预测LVSD患者1年再入院风险,并优于超声参数积分系统。

关键词: 左室功能减低, 高斯过程模型, 超声心动图, 1年再入院率

Abstract: Objective To explore the predictive value of a 1-year readmission risk assessment for patients with chronic left ventricular systolic function(LVSD)by using a Gaussian process model(GPM)and echocardiographic parameters. Methods A total of 290 patients with chronic LVSD were collected, and re-hospitalization within one year was the end point of the study. All patients underwent routine echocardiography to collect data including mitral regurgitation, tricuspid regurgitation, pleural effusion, pericardial effusion, pulmonary artery systolic pressure, left ventricular diameter and left ventricular ejection fraction. The 290 patients were randomly divided into 70% training group(modeling group)and 30% testing group(prediction group). The information interaction analysis on the training dataset was carried out with machine learning algorithms. The importance of each parameter was evaluated, and echocardiographic parameters were merged to establish GPM. Subsequently, the readmission risk of the testing dataset was analyzed using GPM. In addition, the risk of all patients was predicted with the echocardiography parameter scoring system. Finally the two methods were compared with receiver operating characteristic(ROC)curve. Results GPM information interaction analysis showed the weights of echocardiographic parameters were: mitral regurgitation 23.64%, tricuspid regurgitation 22.09%, pleural effusion 16.18%, pericardial effusion 14.36%, pulmonary artery contraction pressure 9.04%, left ventricular diameter 8.86%, and left ventricular ejection fraction 5.83%. Based on GPM and echocardiography parameter scoring system, the area under the ROC curve(AUC)was 83.10%(95% CI: 0.797-0.864), 70.60%(95% CI: 0.647-0.765), respectively. Conclusion GPM based on echocardiographic parameters can well predict the readmission risk of LVSD patients within one year, and it is superior to echocardiography parameter scoring system.

Key words: Left ventricular systolic dysfunction, Gaussian Process Model, Echocardiography, One-year readmission rate

中图分类号: 

  • R541.6
[1] Roger VL, Weston SA, Redfield MM, et al. Trends in heart failure incidence and survival in a community-based population [J]. JAMA, 2004, 292(3): 344-350.
[2] Krittanawong C, Zhang HJ, Wang Z, et al. Artificial intelligence in precision cardiovascular medicine [J]. J Am Coll Cardiol, 2017, 69(21): 2657-2664.
[3] Statnikov A, Wang L. A comprehensive comparison of random forests and support vector machines for microarray-based cancer classification [J]. BMC Bioinformatics, 2008, 9: 319. doi: 10.1186/1471-2105-9-319.
[4] Panahiazar M, Taslimitehrani V, Pereira N. Using EHRs and machine learning for heart failure survival analysis [J]. Stud Health Technol Inform, 2015, 216: 40-44. doi: 10.3233/978-1-61499-564-7-40.
[5] Mortazavi BJ, Downing NS, Bucholz EM, et al. Analysis of machine learning techniques for heart failure readmissions [J]. Circ Cardiovasc Qual Outcomes, 2016, 9(6): 629-640.
[6] 何志昆, 刘光斌, 赵曦晶, 等. 高斯过程回归方法综述[J]. 控制与决策, 2013, 28(8): 1121-1129. HE Zhikun, LIU Guangbin, ZHAO Xijing, et al. Overview of gaussian process regression methods [J]. Control and Decision, 2013, 28(8): 1121-1129.
[7] 翟春晓, 王晓军, 余永明, 等. 心脏超声参数积分系统对慢性左心室收缩功能减低心力衰竭患者1年再入院的风险评估[J]. 中国循环杂志, 2017, 32(12): 1194-1198. ZHAI Chunxiao, WANG Xiaojun, YU Yongming, et al. Risk assessment of echocardiography parameter scoring system for 1 year re-admission in patients with left ventricular systolic dysfunction[J]. China Circulation Journal, 2017, 32(12): 1194-1198.
[8] Seeger M. Gaussian processes for machine learning[J]. Int J Neural Syst, 2004, 14(2): 69-106.
[9] 何显菁. 心力衰竭发病机制、诊断及治疗的研究进展[J]. 实用心脑肺血管病杂志, 2018, 26(7): 1-8. HE Xianjing. Research progress on pathogenesis, diagnosis and treatment of heart failure[J]. Journal of Practical Heart, Brain, Pulmonary and Vascular Diseases, 2018, 26(7): 1-8.
[10] Guidi G, Pollonini L, Dacso CC, et al. A multi-layer monitoring system for clinical management of Congestive Heart Failure[J]. BMC Med Inform Decis Mak, 2015, 15(3): 5-5.
[11] 王晓军, 丛晓, 申晓倩, 等. 采用超声心动图参数分析和预测慢性左心室射血分数降低心力衰竭患者1年内再入院风险[J]. 中国循环杂志, 2016, 31(2): 142-145. WANG Xiaojun, CONG Xiao, SHEN Xiaoqian, et al. Analysis and prediction for 1 year re-admission risk in patients of chronic heart failure with reduced left ventricular ejection fraction by echocardiography[J]. China Circulation Magazine, 2016, 31(2): 142-145.
[12] Giamouzis G, Kalogeropoulos A, Georgiopoulou V, et al. Hospitalization epidemic in patients with heart failure: risk factors, risk prediction, knowledge gaps, and future directions[J]. J Card Fail, 2011, 17(1): 54-75.
[13] 赵玉清, 袁桂莉, 张进顺, 等. 和肽素联合N末端B型利钠肽原评估慢性心力衰竭患者预后的价值[J]. 中国循环杂志, 2014, 29(4): 275-278. ZHAO Yuqing, YUAN Guili, ZHANG Jinshun, et al. Prognostic value assessment of copeptin and NT-proBNP in patients with chronic heart failure[J]. China Circulation Magazine, 2014, 29(4): 275-278.
[14] Levy WC, Mozaffarian D, Linker DT, et al. The Seattle Heart Failure Model: prediction of survival in heart failure [J]. Circulation, 2006, 113(11): 1424-1433.
[15] Giamouzis G, Kalogeropoulos AP, Georgiopoulou VV, et al. Incremental value of renal function in risk prediction with the Seattle Heart Failure Model [J]. Am Heart J, 2009, 157(2): 299-305.
[16] AbouEzzeddine OF, French B, Mirzoyev SA, et al. From statistical significance to clinical relevance: A simple algorithm to integrate brain natriuretic peptide and the Seattle Heart Failure Model for risk stratification in heart failure[J]. J Heart Lung Transplant, 2016, 35(6): 714-721.
[17] Pocock SJ, Ariti CA, McMurray JJ, et al. Predicting survival in heart failure: a risk score based on 39 372 patients from 30 studies[J]. Eur Heart J, 2013, 34(19): 1404-1413.
[18] Senni M, Parrella P, De Maria R, et al. Predicting heart failure outcome from cardiac and comorbid conditions: the 3C-HF score[J]. Int J Cardiol, 2013, 163(2): 206-211.
[19] Obermeyer Z, Emanuel EJ. Predicting the future-big data, machine learning, and clinical medicine [J]. N Engl J Med, 2016, 375(13): 1216-1219.
[20] Shameer K, Johnson KW, Yahi A, et al. Predictive modeling of hospital readmission rates using electronic medical record-wide machine learning: a case-study using mount Sinai heart failure cohort [J]. Pac Symp Biocomput, 2017, 22: 276-287. doi: 10.1142/9789813207813_0027.
[21] 楼浩锋, 张端. 高斯过程下的CMA-ES在医学图像配准中的应用[J]. 计算机科学, 2018, 45(S2): 244-247. LOU Haofeng, ZHANG Duan. Gaussian process assisted CMA-ES application in medical image registration [J]. Computer Science, 2018, 45(S2): 244-247.
[22] Weng SF, Reps J, Kai J, et al. Can machine-learning improve cardiovascular risk prediction using routine clinical data?[J]. PLoS One, 2017, 12(4): e0174944. doi: 10.1371/journal.pone.0174944.
[23] Panahiazar M, Taslimitehrani V, Pereira N, et al. Using EHRs and machine learning for heart failure survival analysis [J]. Stud Health Technol Inform, 2015, 216: 40-44. doi: 10.3233/978-1-61499-564-7-40.
[24] Kennedy EH, Wiitala WL, Hayward RA, et al. Improved cardiovascular risk prediction using nonparametric regression and electronic health record data [J]. Medical care, 2013, 51(3): 251-258.
[25] Ouwerkerk W, Voors AA. Factors influencing the predictive power of models for predicting mortality and/or heart failure hospitalization in patients with heart failure [J]. JACC Heart Fail, 2014, 2(5): 429-436.
[26] Levy WC, Anand IS. Heart failure risk prediction models: what have we learned [J]. JACC Heart Fail, 2014, 2(5): 437-439.
[1] 张淑莹,武晓峰,郭丽敏,乔温,彭洁琼,李大庆. 两种阿霉素心力衰竭模型及心功能进展的评估[J]. 山东大学学报 (医学版), 2020, 58(12): 1-7.
[2] 付开丽1,张学义2,朱萌3,王志浩1,王华1,钟明1,张运1,张薇1. 不同病因心房颤动患者卒中危险的评估[J]. 山东大学学报(医学版), 2013, 51(06): 61-63.
[3] 马丽娜1,姚民强2,王佩显1. 症状期酒精性心肌病患者心脏结构和功能的超声评价[J]. 山东大学学报(医学版), 2011, 49(1): 82-85.
[4] 李婧,王佩显,曹丽,刘丽丽. ACEI对过量饮酒大鼠心脏的保护作用[J]. 山东大学学报(医学版), 2010, 48(6): 22-25.
[5] 赵立健,韩波,张丽萍,张仪,张建军. 骨保护素在诊断川崎病并冠状动脉损害中的应用[J]. 山东大学学报(医学版), 2010, 48(1): 116-119.
[6] 靳有鹏,王玉林,韩波,刘廷亮,张建军,庄建新,韩秀珍,汪翼,马沛然. 不同年龄房间隔缺损封堵术后心脏电生理及几何形态的变化[J]. 山东大学学报(医学版), 2006, 44(8): 806-809.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!