Simple models of cancer growth, MCMC parameter estimation and identifiability
- 2021年4月8日10:00 - 11:00
- カトゥリン・ボシュメン (数理創造プログラム 副プログラムディレクター / Professor, Department of Physics, Ryerson University, Canada)
- via Zoom
I would like to introduce some basic concepts about (very simple) mathematical model of cancer growth, the basic math behind parameter estimation via Markov chain Monte Carlo (MCMC) based on Bayes' theorem, and the different diagnostics you can use to know if the parameters are correctly estimated. I will use a recent example with cancer data in mice.
I think this seminar can be interesting to mathematicians (because of the models and the math behind the parameter estimation, but the math is very basic!), to physicists (especially those that have to do some parameter estimation), and to biologists (the cancer model/data and the parameter estimation). I think it will also be interesting to the information theory and prediction science people. MCMC parameter estimation based on physical models is more valuable in my field than machine learning, so I think those interested in machine learning but maybe are not so familiar with MCMC should join to consider them as an alternative approach in certain contexts.
*Please refer to the email to get access to the Zoom meeting room.