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 (9): 71-76.doi: 10.6040/j.issn.1671-7554.0.2020.0555

• 公共卫生与管理学 • 上一篇    下一篇

利用数据库数据采用联合模型动态预测312例肝硬化患者预后的观察分析

肖宇飞,冯佳宁,王晓璇,毛倩,石福艳,王素珍   

  1. 潍坊医学院公共卫生学院, 山东 潍坊 261053
  • 出版日期:2020-09-10 发布日期:2020-08-30
  • 通讯作者: 石福艳. E-mail:shifuyan1@126.com; 王素珍. E-mail:wangsz@wfmc.edu.cn
  • 基金资助:
    国家自然科学基金(81803337;81872719);山东省高等学校青创人才引育计划(No2019-6-156,Lu-Jiao);山东省自然科学基金(ZR2019MH034);国家统计局课题(2018LY79);潍坊医学院博士启动基金(2017BSQD51)

Prediction of the prognosis of 312 cirrhosis patients using the joint models and database

XIAO Yufei, FENG Jianing, WANG Xiaoxuan, MAO Qian, SHI Fuyan, WANG Suzhen   

  1. School of Public Health, Weifang Medical University, Weifang 261053, Shandong, China
  • Online:2020-09-10 Published:2020-08-30

PDF (PC)

451

摘要： 目的 分析在标准联合模型基础上添加纵向监测变量斜率项后在纵向生存数据中的动态预测效能。 方法 基于R 3.6.2软件JM包中的原发性胆汁肝硬化(PBC2)数据集,分别构建标准联合模型和时变斜率参数化联合模型,分析两种联合模型的拟合结果,对比两种联合模型对肝硬化患者生存率的预测效能,对患者进行个体生存率的动态预测。 结果 标准联合模型结果显示:log(serBilir)每纵向增加一个单位,患者发生死亡风险增加3.852 0倍(95%CI:3.152 5~4.706 3)。时变斜率参数化的联合模型结果显示:log(serBilir)每纵向增加一个单位,患者发生死亡风险增加3.135 2倍(95%CI:2.433 2~4.039 8)。纵向轨迹斜率每增加一个单位,患者死亡风险就增加17.431 9倍(95%CI:2.399 6~126.633 9),log(serBilir)纵向轨迹的斜率与肝硬化患者的死亡风险高度相关(P<0.000 1)。比较两个模型的动态预测效能,两个联合模型均表现出患者未来4年的生存率比未来2年精度要高。而时变斜率参数化联合模型的ROC曲线下面积(AUC)范围(0.714 2~0.831 1)比标准联合模型AUC范围(0.645 0~0.836 1)小,预测性能更稳定。对随机1例患者预测4年后的生存率为0.850 3(95%CI:0.481 0~0.963 9)。 结论 标准联合模型和时变斜率参数化联合模型均可用于分析纵向生存资料中纵向监测指标与生存结局之间的关联性,而且时变斜率参数化联合模型在满足标准联合模型的功能同时考虑了纵向监测变量轨迹斜率对生存结局的影响,具有较好的拟合优度,比标准联合模型有较高的预测性能。

关键词: 联合模型, 斜率参数化, 纵向数据, 生存分析, 预测

Abstract: Objective To explore the dynamic prediction efficiency of longitudinal survival data after slope term of longitudinal monitoring variables was added on the standard joint model. Methods Based on the dataset of primary biliary cirrhosis(PBC2)in the R 3.6.2 software of JM package, a standard joint model and a time-varying slope parameterized joint model were constructed, whose fitting results were analyzed. Their prediction efficiency on the survival rate of cirrhosis patients were compared and the individual survival rate was predicted. Results The standard joint model showed that with each unit of log(serBilir)increased longitudinally, the risk of death increased by 3.852 0 times(95%CI: 3.152 5-4.706 3). The time-varying slope parameterized joint model showed that with each unit of log(serBilir)increased longitudinally, the risk of death increased by 3.135 2 times(95%CI: 2.433 2-4.039 8). If the slope of the longitudinal trajectory increased by one unit, the risk of death increased by 17.431 9 times(95%CI: 2.399 6-126.633 9). The slope of log(serBilir)longitudinal trajectory was highly correlated with the risk of death(P<0.000 1). Both models were more accurate in predicting the 4-year survival rate than predicting the 2-year survival rate. The area under the ROC curve(AUC)of the time-varying slope parameterized joint model(0.714 2-0.831 1)was smaller than that of the standard joint model(0.645 0-0.836 1), and the prediction performance was more stable. The time-varying slope parameterized joint model predicted the 4-year survival rate was 0.850 3(95%CI: 0.481 0-0.963 9)for 1 randomized patient. Conclusion Both the standard joint model and time-varying slope parameterized joint model can be used to analyze the correlation between longitudinal monitoring indexes and survival outcomes in longitudinal survival data. The time-varying slope parameterized joint model not only has the function of the standard joint model, but also takes into account the influence of the trajectory slope of longitudinal monitoring variables on survival outcomes. Therefore, it has higher prediction efficiency than the standard joint model.

Key words: Standard joint model, Slope parameterization, Longitudinal data, Survival analysis, Prediction

中图分类号: 

  • R18
[1] 赵悦. 纵向数据与生存数据联合模型的变量选择[D]. 长春: 东北师范大学, 2019.
[2] 张慧敏. 基于联合模型的纵向和生存数据统计方法探讨[D]. 南京: 东南大学, 2019.
[3] 孙申. 基于一组肝硬化临床实验数据的两个纵向和生存数据的联合模型及其比较研究[D]. 昆明: 云南师范大学, 2019.
[4] Wulfsohn MS, Tsiatis AA. A joint model for survival and longitudinal data measured with error [J]. Biometrics, 1997, 53(1): 330-339.
[5] Tsiatis AA, Degruttola V, Wulfsohn MS. Modeling the relationship of survival to longitudinal data measured with error: applications to survival and CD4 counts in patients with AIDS [J]. J Am Stat Assoc, 1995, 90(429):27-37.
[6] 范永君, 刘启贵, 刘珂华, 等. 尿酸的动态变化与代谢综合征的关联研究[J]. 中国卫生统计, 2019, 36(4): 507-510. FAN Yongjun, LIU Qigui, LIU Kehua, et al. The correlation between dynamic changes of uric acid and metabolic syndrome [J]. Chinese Journal of Health Statistics, 2019, 36(4): 507-510.
[7] Rizopoulos D, Murawska M, Andrinopoulou ER, et al. Dynamic predictions with time-dependent covariates in survival analysis using joint modeling and landmarking [J]. Biom J, 2017, 59(6): 1261-1276.
[8] 严国义. 纵向数据与生存数据的半参数联合模型研究[D]. 武汉: 武汉大学, 2013.
[9] Rizopoulos D, Hatfield LA, Carlin BP, et al. Combining dynamic predictions from joint models for longitudinal and time-to-event data using Bayesian model averaging [J]. J Am Stat Assoc, 2014, 109(508):1-29.
[10] Huang Y, Yan C, Xing D, et al. Jointly modeling event time and skewed-longitudinal data with missing response and mismeasured covariate for AIDS studies [J]. J Biopharm Stat, 2015, 25(4): 670-694.
[11] 王策. 纵向和生存数据的一个联合建模方法研究及其在艾滋病疗效评估中的应用[D]. 昆明: 云南师范大学, 2015.
[12] 王旭霞. 基于联合模型的认知功能和社会功能对阿尔茨海默病风险预测研究[D]. 太原:山西医科大学, 2019.
[13] Li K, Luo S. Dynamic predictions in Bayesian functional joint models for longitudinal and time-to-event data: An application to Alzheimers disease [J]. Stat Methods Med Res, 2019, 28(2): 327-342.
[14] Rizopoulos D, Biometrics. Dynamic predictions and prospective accuracy in joint models for longitudinal and time-to-event data [J]. Biometrics, 2011,67(3): 819-829.
[15] 王一茸, 韦程东, 岑泰林, 等. 基于纵向生存数据联合模型的肝硬化数据应用研究[J]. 广西师范学院学报(自然科学版), 2018, 35(2): 31-36. WANG Yirong, WEI Chengdong, CEN Tailin, et al. Liver cirrhosis data application research based on the joint model of longitudinal data and survival[J]. Journal of Nanning Teachers Education University(Natural Science Edition), 2018, 35(2): 31-36.
[16] 孙宏鸽, 刘启贵, 王晓蓉, 等. 基于联合模型的谷丙和谷草转氨酶动态变化与代谢综合征的关联研究[J]. 中国卫生统计, 2018, 35(2): 181-185. SUN Hongge, LIU Qigui, WANG Xiaorong, et al. A longitudinal analysis on the relationship between dynamic changes of alanine aminotransferase and a longitudinal analysis on the relationship between dynamic changes of alanine aminotransferase and metabolic syndrome based on the joint model[J]. Chinese Journal of Health Statistics, 2018, 35(2): 181-185.
[17] 李淞淋, 魏戌, 易丹辉. 基于纵向和生存时间结局的联合模型方法及应用[J]. 世界科学技术-中医药现代化, 2019, 21(3): 366-373. LI Songlin, WEI Xu, YI Danhui. Methodology and application of joint models based on longitudinal endpoints and time to events [J]. Modernization of Traditional Chinese Medicine and Materia Medica-World Science and Technology, 2019, 21(3): 366-373.
[18] Ibrahim JG, Chu H, Chen LM. Basic concepts and methods for joint models of longitudinal and survival data [J]. J Clin Oncol, 2010, 28(16): 2796-2801.
[19] Garre FG, Zwinderman AH, Geskus RB, et al. A joint latent class changepoint model to improve the prediction of time to graft failure [J]. J R Stat Soc Series B Stat Methodol, 2008, 171(1): 299-308.
[20] 池铭. 纵向项目响应与有治愈的生存时间的联合建模[D]. 大连: 大连理工大学, 2019.
[21] 翟映红, 陈琪, 韩贺东, 等. 联合模型介绍及在医学研究中的应用[J]. 中华流行病学杂志, 2019, 40(11):1456-1460.
[1] 吴强,何泽鲲,刘琚,崔晓萌,孙双,石伟. 基于机器学习的脑胶质瘤多模态影像分析[J]. 山东大学学报 (医学版), 2020, 58(8): 81-87.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
版权所有 © 2018 《山东大学学报(医学版)》编辑部　
地址:山东大学科技期刊社（济南市历城区山大南路27号山东大学中心校区明德楼B座721室） 电话: 0531-88366918 E-mail: xbyxb@sdu.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发