SIMULATING BREAST CANCER IN WISCONSIN

NCI癌症干预和监测建模网络(CISNET)的目的是促进模拟建模作为一种工具，与国家的癌症监测系统相结合，帮助解释观察到的癌症发病率和死亡率的变化。威斯康星州提供了一个独特的人口实验室来开发和测试乳腺癌模拟模型，以满足这一目标。我们建议威斯康星大学的模拟和统计专家、监测和流行病学专家以及威斯康星州的癌症报告系统合作，研究模拟建模的使用，以更好地了解乳腺癌流行病学的趋势，并为此加强模拟建模的使用。该项目将更新和扩展先前开发的模拟威斯康星州乳腺癌年龄和阶段特定发病率和年龄特定死亡率的模型。该模型是在1992-93年开发和验证的，并用于解释1982-1992年威斯康星州乳腺癌的趋势。我们将把宏观模拟模型重新编程为离散事件微模拟，更新输入以考虑自1992年以来的人口统计、乳腺癌检测和治疗变化。修改还将纳入关于环境、行为和遗传危险因素的新的流行病学知识，并特别注意解释关于原位乳腺癌的新知识。在做出这些改变时，我们将探索使用贝叶斯估计技术来根据基于时间的监视数据来拟合模拟参数。该模型将与地理编码的监测数据放在一个地理信息分析平台上，并设计出系统比较模型预测和监测数据之间的区域差异的技术。与其他受资助者合作，将探讨模型开发、验证和使用方面的问题。

The purpose of the NCI Cancer Intervention and Surveillance Modeling Network (CISNET) is to promote simulation modeling as a tool that, in conjunction with the nation's cancer surveillance systems, can help to explain observed changes in cancer incidence and mortality. Wisconsin offers a unique population laboratory to develop and test breast cancer simulation models to meet this goal. We propose a collaboration among simulation and statistics experts, and surveillance and epidemiology experts at the University of Wisconsin, and the state of Wisconsin's Cancer Reporting System to study the use of simulation modeling to better understand trends in breast cancer epidemiology and to enhance the use of simulation modeling for this purpose. This project will update and extend a previously developed model simulating breast cancer age- and stage-specific incidence and age-specific mortality in Wisconsin. The model was developed and validated in 1992-93 and was used to explain breast cancer trends in Wisconsin from 1982-1992. We will reprogram the macrosimulation model as a discrete event microsimulation, updating inputs to account for demographic, and breast cancer detection and treatment changes since 1992. Modifications will also incorporate new epidemiologic knowledge about environmental, behavioral, and genetic risk factors, with special attention to account for new knowledge about breast cancer in situ. In making these changes we will explore use of Bayesian estimation techniques for fitting simulation parameters in light of time- based surveillance data. The model will be placed on a geographic information analysis platform along with geo-coded surveillance data and devise techniques for systematic comparison of regional variations between model predictions and surveillance data. Collaboration with other CISNET grantees will explore issues in model development, validation, and use.