Applications of Multi-Criteria Optimization (AMCO) to Cancer Simulation Modeling

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Cancer screening programs are increasingly evaluated with simulation models because they allow health policy makers to consider scenarios that could not be evaluated by randomized clinical trials for practical, financial or ethical reasons. However, few of these models employ rigorous mathematical methods for model calibration. Calibration of cancer screening simulation models to existing clinical data is vital to accurate model prediction. The applicant's immediate goal is to adapt, extend, and promote the use of multi-criteria optimization techniques to improve the calibration of simulation models for cancer screening policy prediction and planning. The applicant, Chung Yin Kong, PhD, is a senior scientist at the Massachusetts General Hospital's Institute for Technology Assessment (ITA) and an instructor at Harvard Medical School. He is trained in Physics (BS) and Polymer Science and Engineering (PhD). This proposed research is tailored to utilize his computer modeling background in physical science as well as the numerous simulation projects at the ITA to test his hypotheses for improving the design and construction of cancer screening models with multi-criteria optimization techniques. The specific aims of the research plan are: (1) to adapt multi-criteria optimization to provide automated procedures for model calibration. As an example, optimization algorithms will be applied to and evaluated with two existing microsimulation models at the ITA: the Lung Cancer Policy Model (LCPM) and the Simulation Model of Colorectal Cancer (SimCRC) model; (2) to extend the use of multi-criteria optimization techniques to aid the design of the underlying cancer biology components in the models and to improve computational speed; (3) to promote the use of multi-criteria optimization techniques among cancer screening modelers. The experience of adapting and extending these techniques will be developed into a calibration platform with instructional diagrams, tutorials, and software modules, which will be distributed on the Internet and at scientific conferences. The end results of the proposed project will improve the speed of both the calibration process and the simulation models themselves. The proposed training plan includes mentoring, coursework, and career development activities preparing him to undertake the proposed research and to fully-transition into the field of cancer simulation modeling. The research and training of this proposed project will be performed under the mentorship of Dr. G. Scott Gazelle, an internationally known expert in cancer outcome research and decision analysis science. The applicant's long term career goal is to become a leader in developing state-of-the-art simulation methods for disease modeling. This award will advance the applicant's academic career and help him to achieve his goal to be a productive, independent investigator. PUBLIC HEALTH RELEVANCE: This research is relevant to public health because it improves the accuracy of simulation models for cancer screening policy prediction and planning.