论文摘要
本文探讨了生存分析模型的理论基础、模型拟合、变量选择和假设检验,并用实例说明了如何将该模型用于确定接触蚊子寄生虫和健康危险的关系。疟疾案例是加纳地区四家医院在四个月内的从病人住院到最终状态(死亡或再次检查)的数据,考虑偶然因素(暴露于疟疾寄生虫的水平),并用生存分析建模技术分析。在取样的1793个病人中,405代表22.6%正在使用的驱虫蚊帐同时大多数病人(77.4%)用了其他类型的蚊帐,例如窗纱和普通蚊帐。它建立在在那些使用驱虫蚊帐的患者中,在四个月的研究内有16%的人死亡,而84%的幸存;但在相同的研究内,在未用驱虫蚊帐的患者当中,23%的人死亡77%的人幸存。男性和女性的生存期大致相同;这意味着,性别对于那些死于疟疾的病人的影响不是很显著。危险比率HR(t|exp osed group,less exp osed group)=1.5,表明暴露组比少暴露组的疟疾死亡风险高。Kaplan-Meier曲线显示有驱虫蚊帐(ITNs,用来减少暴露疟疾寄生虫)的病人比没有驱虫蚊帐的病人预计要活得好和长久。图形检验指出两个暴露水平下的生物医学数据平行;对数图形的平行是一个石蕊检验,用来评估相对危险(PH)假设。生存概率生命表同时也表明,愿意使用驱虫蚊帐作为阻止寄生虫感染手段的患者比不愿意使用的患者要活的好和长久。假设检验的结果是意义深远的。首先,从对数秩的检验结果(p值=0.0020<0.0500)可以得出,如果我们研究了所有人口,并检测了两组的疟疾寄生虫暴露水平和健康危险的关系,则少暴露组和暴露组的生存曲线不同,少暴露组比暴露组有更长的健康生存时间。其次,我们注意到如果整个被诊断出患有的疟疾的患者被纳入研究范围,驱虫蚊帐使用组和非使用组存在显著差异;再次,由似然比试验的p值(0.0035<0.0500)可以得出,居民用来防止暴露于寄生虫的方法是有效的,结果表明,使用驱虫蚊帐做为减少寄生虫感染的手段是很有意义的。第四,从瓦尔德的统计资料中可以看到,在调整了性别变量以后,这种被居民用来防止寄生虫感染的方法是很有效的。最后,从Mantel-Cox残余试验可以得到,在显著性水准为0.05和0.01时,失败的次数排名和Schoenfeld残余的关联是很重要的。这一结果足以证明驱虫蚊帐在防止疟疾死亡的风险上是有效的。研究结果表明需要进一步呼吁卫生保健专业人员和个人要采纳驱虫蚊帐作为疟疾祸害的手段。
论文目录
摘要ABSTRACTTABLE OF CONTENTLIST OF TABLESLIST OF FIGURESCHAPTER ONE:INTRODUCTION1.1 Background to the study1.2 Problem statement1.3 Rational of the study1.4 Study objectives1.5 Research questions1.6 Formulation of research hypothesis1.7 Outline of chaptersCHAPTER TWO:LITERATURE REVIEW2.0 Introduction2.1 Background of district under study2.2 The disease malaria and its symptoms2.3 Distribution of endemic malaria cases by countries2.4 Transmission of malaria2.5 Insecticide treated mosquito nets(ITNs)2.6 Usage of insecticide treated nets2.7 Ownership and usage of insecticide treated nets in Ghana2.8 Research/interventions on ITNs and malaria control programs in Ghana2.9 Economic and social burdens of malaria on the societyCHAPTER THREE:SURVIVAL ANALYSIS MODELS3.0 Introduction3.1 Survival Analysis3.2 Time scales,origins and terminations3.3 Describing the Survival time data3.4 Importance of survival analysis3.5 Goals of survival analysis3.6 Concept of censoring3.7 Truncation3.8 Likelihood construction for censored and truncated data3.9 Time-dependent covariates3.10 Survival analysis distribution3.11 Non-parametric methods of estimation3.12 Life tables3.13 Product limit estimator3.14 Parametric methods of estimation3.15 Failure time distributions3.16 Regression models3.17 Accelerated failure time model(AFT)3.18 Method for assessing the proportional hazard(PH) assumption3.19 Maximum likelihood estimation3.20 Hazard models3.21 Hazard ratio3.22 Developing the Cox proportional hazard model3.23 Stratified Cox Model3.24 Graphical approach to the log-log plot3.25 Likelihood ratio test3.26 Log-rank test for the exposed and less exposed groups(LRT)3.27 Stratified log-rank testCHAPTER FOUR:METHODOLOGY4.0 Introduction4.1 Target population4.2 Coding of data4.3 Data analysis4.4 Pictorial survival model for biomedical dataCHAPTER FIVE:RESULTS OF ANALYSIS5.0 Introduction5.1 Comparing survival experiences of users and non-users of ITNs5.2 Comparing the hazards rates of users and non-users of ITNs5.3 Comparing the survival experiences of males and females5.4 Assessing the proportional hazards model5.5 Pictorial representation of hazards of malaria death5.6 Running of stratified Cox model5.7 Computer results of Cox proportional hazard model5.8 Hypothesis testingCHAPTER SIX:SUMMARY,DISCUSSIONS AND CONCLUSIONS6.0 Introduction6.1 Summary6.2 Discussion6.3 ConclusionACKNOWLEDGEMENTREFERENCESAPPENDICESAPPENDIX 1:List of published papers during the study for a degree
相关论文文献
标签:生物医学数据论文; 暴露到环境论文; 驱虫蚊帐论文;