Instructor: Prof. Nathaniel Osgood

Email: osgood@cs.usask.ca

Course Dates: September 1st - December 7th, 2022

Registration Deadline: September 8th, 2022

Lecture Times: Tuesday & Thursday | 10:30 am - 12:00 pm (CST)

Office Hours: Tuesday & Thursday | 12:00 - 1:00 pm (CST) 

Cost: Free for members (students, postdocs, etc) of Canadian Unviersities - $500 CAD for students outside of Canada 

Course Overview: 

This course will offer a systematic introduction to the theory and practice of agent-based modeling, with a emphasis on infectious disease modeling. Course lectures will cover the breadth of the agent-based modeling process, including model conceptualization and scoping, model architecture, design and formulation, model implementation, testing for verification and validation, calibration and sensitivity analysis, and scenario evaluations. Topics discussed include mechanisms for specifying discrete, continuous and relational aspects of model state and their evolution, alternative prevalent model architectures, representing and the structure and dynamics of diverse classes of networks, mechanisms for agent-agent and agent-environment interaction, model observation processes, meta-population models, and model performance and scaling with heterogeneity and population size. Lectures will further examine alternative models of time and space, spatially and geograhically explicit models and integration with geographic information system data, representing situated perception, and behaviours and preferences in agent-based modeling, and model integration with big-data. Several key lectures will focus on modeling process issues, including the imperative of pursuing agent-based modeling processes in an incremental fashion, and the importance of the YAGNI principle, model docking, and agile modeling approaches. Continuing the emphasis of this process module and reflective of the strong importance of involving those with lived experience of health conditions and access barriers for grounded understanding of many health needs, the couse will also discuss the use of agent-based models in participatory modeling, their use as story-telling tools, and characterize the distinctive benefits secured by participatory modeling processes employing agent-based modeling. Discussions of hybrid models will emphasize the complementary insights to be secured when agent-based and hybrid modeling is combined with other types of dynamic modeling, with a particular emphasis on a handful of hybrid strategies that have conferred repeated value across past and existing projects. As time allows, the couse will further discuss model state space, coupling and dimensionality and identifiability, the great value of agent-based models when used as souces of synthetic data for machine learning and statistical analysis, and the multi-faceted nature of emergence.