Empirical Likelihood Bayes Analysis of Quantile Regression Model

描述(由申请人提供)： 这项研究的目的是使用创新的贝叶斯分位数回归(QR)方法和删失QR模型重新分析儿童癌症幸存者研究(CCSS)和癌症和白血病B组(CALGB)研究9082个数据。CCSS和CALGB研究9082分别涉及癌症流行病学的两个重要问题，即识别和评估可改变疾病的风险因素，以及权衡更具侵袭性的癌症治疗与侵袭性较低的癌症治疗。这些研究的初步分析受到研究设计或统计分析方法选择的限制，未能充分回答各自的研究问题。通过适当地解决这些限制因素，重新分析这些研究是很有必要的。我们将使用一种创新的贝叶斯(经审查)QR方法。它在贝叶斯分析框架内用经验似然取代似然，并进行(经审查的)QR分析。这将是第一种也是唯一一种关于审查的QR的方法，也是少数几个关于未审查的QR的方法之一，它包含了充分重新分析CCSS和CALGB研究9082数据所必需的所有特征。它还将为许多其他癌症流行病学研究提供另一种分析工具，在这些研究中，标准回归或生存数据分析方法失败或被认为不充分。这项研究将分别提高我们对儿童急性淋巴细胞性白血病(ALL)幸存者肥胖相关未来健康风险、头颅放射治疗的作用以及高剂量化疗(结合自体骨髓移植)对高危乳腺癌女性的相对益处的理解。因此，这些知识将导致更好地设计治疗方案和干预策略，从而提高儿童癌症和高危乳腺癌患者群体的存活率，并将对健康的有害影响降至最低。 公共卫生相关性： 本研究的目的是使用创新的贝叶斯分位数回归方法和删失分位数回归模型重新分析儿童癌症幸存者研究(CCSS)和癌症与白血病B组(CALGB)研究9082。拟议的分析将分别提高我们对儿童急性淋巴细胞白血病(ALL)幸存者的肥胖相关未来健康风险、头颅放射治疗的作用以及高剂量化疗和自体骨髓移植在高危乳腺癌女性中的相对益处的理解。

DESCRIPTION (provided by applicant): The purpose of this study is reanalyzing the Childhood Cancer Survivors Study (CCSS) and the Cancer and Leukemia Group B (CALGB) study 9082 data using an innovative Bayesian method for quantile regression (QR) and censored QR model. The CCSS and the CALGB study 9082 are concerned with two important questions of cancer epidemiology, respectively, identifying and assessing risk factors for modifiable disease and weighing more aggressive cancer treatment against less aggressive one. The primary analyses of the studies were constrained by the study design or the choice of statistical analysis method and fell short of adequately answering their respective research questions. Reanalyzing these studies is well warranted by appropriately addressing such constraints. We will use an innovative Bayesian (censored) QR method. It replaces the likelihood with empirical likelihood within the Bayesian analytic framework and conducts (censored) QR analysis. It will be a method, the first and only regarding the censored QR and one among a few regarding the uncensored QR, that accommodate all the features that are essential to adequately reanalyze the CCSS and the CALGB study 9082 data. It will also provide an alternative analytic tool for many other cancer epidemiology studies where standard regression or survival data analysis methods fail or are deemed insufficient. This study will improve our understanding of, respectively, obesity related future health risks in the childhood acute lymphoblastic leukemia (ALL) survivors and the role of cranial radiotherapy, and of the relative benefit of the high dose chemotherapy (with autologous bone marrow transplantation) in women with high risk breast cancer. This knowledge will consequently lead to better designed treatment protocols and intervention strategies that will increase survival and minimize harmful health effects in pediatric cancer and high risk breast cancer patient populations. PUBLIC HEALTH RELEVANCE: The purpose of this study is reanalyzing the Childhood Cancer Survivors Study (CCSS) and the Cancer and Leukemia Group B (CALGB) study 9082 by using an innovative Bayesian method for quantile regression and censored quantile regression model. The proposed analyses will improve our understanding of, respectively, obesity related future health risks in the childhood acute lymphoblastic leukemia (ALL) survivors and the role of cranial radiotherapy, and of the relative benefit of high dose chemotherapy with autologous bone marrow transplantation in women with high risk breast cancer.