English Version

首页» 学术科研» 讲座信息

4月18日工业工程与管理系——On the Cutting Edge of Statistical Research:Focusing on Cancer Survival Modeling and

讲座题目On the Cutting Edge of Statistical Research: Focusing on Cancer Survival Modeling and Prediction for Computer Experiments

报告人Gang Han

时 间418日(周一)上午1000-1200
地 点:北京大学力学楼434


The integration and collaboration of computer science, health sciences, and statistics have been developed and implemented extensively. Given this trend, understanding of the importance and necessity of statistical science is crucial for scholars in all research fields. This talk will first briefly introduce some areas of advances in statistical research. The discussion following will mainly focus on two areas on the forefront of statistical research: one, an improved piecewise exponential model for cancer survival; two, the analysis of complex computer models having qualitative and quantitative inputs.
    1. In medical research, statistical models for survival data are typically non- or semi-parametric, e.g., the Kaplan-Meier curve. A major constraint of the existing parametric models is the lack of flexibility due to distribution assumptions. A flexible and parsimonious piecewise exponential model is introduced in this talk. Such modeling provides an additional descriptive tool in understanding differences in patient response and can also serve as a validation tool for exponential failure. An application in a lung cancer study is presented.
    2. Gaussian Stochastic Process models are a popular way to model output from computer experiments. Unfortunately, the correlation functions typically used in these models assume inputs are quantitative. This talk introduces a method for handling both quantitative and qualitative inputs. The proposed method builds on an ANOVA-type decomposition of the response variable and a Bayesian hierarchical approach with an empirical prior. It improves some existing approaches when there is a common underlying shape in the response for each level of the qualitative variables. An application is demonstrated in a hip prosthetic device finite element analysis computer simulation model.


Dr. Han is a member in the Biostatistics Department at the Moffitt Cancer Center. He received his Ph.D. in Statistics from the Ohio State University in 2008. His research efforts have been in Statistics, Computer Science, and their applications to Biomedical Research and Bioinformatics. He has been working on the design and analysis of complex computer models since 2004. He developed statistical approaches for modeling the output from complex computer codes having quantitative and quantitative inputs, as well as the simultaneous calibration and tuning for computer models. These approaches were motivated by the prosthesis design at Cornell University as well as the Hospital for Special Surgery, New York, and are expected to be applied in the near future. His work at the Moffitt Cancer Center includes methodological research such as innovative modeling and comparing cancer survivals in clinical trials and national registration data base. Also, His work involves applications such as modeling Epidemiology data, and statistical learning in cancer detection, and building comparative effectiveness research infrastructure.