报告题目：Challenges in real-world modeling: nonindentiability issues and model selection

报告时间：2020年4月25日（周六）上午10:00—10:40

报 告 人：Michael Li教授

报告地点：Zoom云会议（ID：932 1144 3567, 密码：nanxinda60）

报告摘要：The disease modeling communities around the world have been actively participated in model projections for COVID-19 epidemics. Some of the lessons we have learned during this process are common to other modeling exercises of real-world diseases that use public health data: models are used to interpret the data and make short-term predictions that aim to inform public health policy. Traditional training in theoretical mathematical epidemiology focuses on long-term behaviours using tools of stability and bifurcation analysis. The approach is often to incorporate more epidemiological complexities into models and look for more interesting dynamics.

The goal of theoretical modeling is to categorizes the entire parameter space into different regions of qualitatively distinct dynamics. Real-world disease modeling such as modeling the COVID-19 epidemics has a very different nature: (1) the problem is more quantitative and finite time, so the stability concept and analysis are often not applicable; (2) we are looking for a set or a range of parameter values for which the model best describes the observed data with noises, so that bifurcation analysis is often not the goal; (3) The same model with different parameter values can give very different projections, so a model should be judged together with its specific set of parameter values rather than only by its model structure; and (4) the best model may not be the most realistic, so it is important to select the best model among a collection with a varying degree of complexity.

In this talk, I focus on two fundamental challenges to real-world modeling of infectious diseases. The first one is why should we and how do we select the best-performing model among a collection of nested models with different complexity. The second is the nonidentifiablity problem in which infinitely many sets of parameter values can produce the same representation of the observed data while giving very different future predictions. These challenges need to be rigorously addressed by the modeling community for disease models to be a reliable tool for public health research.

欢迎广大师生踊跃参加！

 数学与统计学院

2020年4月23日

附：专家简介

Michael Li is a Professor of Mathematics at the University of Alberta, Canada. His research interests expertise are in the theory and applications mathematical modeling of infectious diseases in general, and of HIV, influenza and Tuberculosis in particular, viral dynamics and immune responses dynamics to viral infections including HIV-1 and HTLV-1. Professor Li obtained his PhD in Applied Mathematics at the University of Alberta and did his postdoctoral training at the University of Montreal and Georgia Institute of Technology. He has been a faculty member at the University of Alberta since 2000, where he actively collaborate with research groups in the faculty of medicine and at the Alberta Ministry of Health on modeling research in health and public health sciences.