BIOMATHEMATICAL APPROACHES TO CANCER

相对风险回归模型正越来越多地应用于 癌症流行病学及临床资料分析。这个 这些模型在中小尺度下的统计特性 数据集并没有得到很好的理解。这笔赠款的第一部分 申请将调查推理的程序、方法 模型批评和验证，以及计算方面的考虑 在使用一般相对风险回归模型进行分析时 流行病学或临床数据。这其中的一个重要特征是 研究将系统地使用新开发的工具 关于统计学中的微分几何。第二部分： 格兰特提案将解决癌症风险量化的问题 评估。一个先前发展的随机两阶段模型 致癌作用将被延长，以允许将时间- 相依协变量。此模型的属性将为 在风险评估的背景下进行调查。

Relative risk regression models are being increasingly use in the analysis of cancer epidemiologic and clinical date. The statistical properties of these models in small to moderate-sized data sets are not well understood. The first part of this grant application will investigate procedures for inference, methods of model criticism and validation, and computational considerations when using general relative risk regression models for analysis of epidemiologic or clinical data. An important feature of this research will be the systematic use of the newly developed tools of differential geometry in statistics. The second part of the grant proposal will address the problem of quantitative cancer risk assessment. A previously developed stochastic two-stage model for carcinogenesis will be extended to allow the incorporation of time- dependent covariates. The properties of this model will be investigated within the context of risk assessment.