Dr. Nathaniel D. Osgood

Dr. Nathaniel D. Osgood is a Professor in the Department of Computer Science and Associate Faculty in the Department of Community Health & Epidemiology at the University of Saskatchewan. His research is focused on providing cross-linked simulation, ubiquitous sensing, and machine learning tools to inform understanding of population health trends and health policy tradeoffs. His applications work has addressed challenges in the communicable, zoonotic, environmental, and chronic disease areas.  Dr. Osgood is further the co-creator of two novel mobile sensor-based epidemiological monitoring systems, most recently the Google Android- and iPhone-based iEpi (now Ethica Health) mobile epidemiological monitoring systems.  He has additionally contributed innovations to improve dynamic modeling quality and efficiency, introduced novel techniques hybridizing multiple simulation approaches and simulation models with decision analysis tools, and which leverage such models using data gathered from wireless epidemiological monitoring systems.  Dr. Osgood has led many international courses in simulation modeling and health around the world, and his online videos on the subject attract thousands of views per month.  Prior to joining the U of S faculty, he graduated from MIT with a PhD in Computer Science in 1999, served as a Senior Lecturer at MIT and worked for a number of years in a variety of academic, consulting and industry positions.

Teaching Assistants

The course will be staffed with a broad set of graduate-level teaching assistants, who will provide assistance both during the tutorial sessions and during the open times and post-tutorial brainstorming sessions.  To better address the questions of participants from a wide variety of backgrounds, the teaching assistants will be drawn from both health science and technical backgrounds.

Teaching Style

The tutorial will employ a hands-­on teaching style where participants will learn concepts while interactively exploring existing modeling software, model structure and model results. The instructor will use examples to highlight unique features of agent­based models, and the types of questions on which they offer particular advantage, and limitations of such models. Registered participants will be given access to examples prior to the tutorial, and will be provided usable working copies of AnyLogic for the duration of the tutorial.

Additional times will be set aside for experimenting with the modeling package, pursuing exercises, raising points for clarification, and brainstorming with modelers and TAs concerning possible modeling projects.

Tutorial Contents

The tutorial will consist of a set of plenary sessions covering in which the participants will build example models as guided by the instructor (and with a pool of teaching assistants circulating to assist). Such exercises aid the participant in exploring ways of applying the tool to investigate particular types of health issues, and building strengths in use of the platform. A large fraction of the event will be spent in sessions where participants will have opportunity to pursue their own modeling projects in teams or alone, or may opt to participate in parallel sessions covering more detailed or more technically demanding material.

Optional Sessions

Additional background sections will be offered for practitioners interested in seeking broader background or deeper understanding of some modeling practices and supporting technologies. The particular set of sessions to be offered will depend on the balance student interest expressed in the candidate sessions.

While we cannot guarantee that all topics listed below will be covered during the bootcamp, dedicated sessions will cover material of interest to participants in a prioritized fashion, and videos will be made available for those topics which could not be included within the bootcamp.

  • Motivation for systems science methods in general, and agent­based modeling in particular
  • Best practices for model building
  • Model performance
    • Common performance vulnerabilities
    • Measuring performance using profilers
    • Effective strategy for improving model performance
  • Helpful bits of Java for AnyLogic users who lack programming exposure but are interested in securing added flexibility when creating models, and deeper understanding of model plumbing.
    • Basic introduction to Java
    • Methods & functions
    • Classes & objects
    • Events
    • Types
    • Expressions
    • Statements
    • Using code in external Java libraries
    • Capturing hierarchies of related agents or resources: Subtyping and subclassing
  • Simulating the flow and interaction of agents and resources in facilities: Discrete event modeling and visualization in AnyLogic
  • Discussion on use of agent­based models for study planning, including evaluation of robustness of study design or statistical measures.

Optional post-tutorial modelling brainstorm session

Optional post-tutorial modelling brainstorm session

The instructor and teaching assistants will be available to participants for a day of modeling brainstorming. This session will provide participants a chance to integrate and assimilate teachings from the tutorial session, and discuss application to their area of interest.


By the conclusion of the tutorial, each participant will be provided with the following:

  • Slides
  • Exercises
  • Models built by the participant during in­class exercises
  • Example models provided to participants
  • Models built by the participant with TA & instructor guidance
  • References to more extensive learning opportunities (full courses, articles, books and model libraries)
  • Guidance on accessing videos of course contents and related online material
  • Custom­built tools to ease model insight and debugging


Why Systems Science?

Despite widespread progress in prolonging life over the past century, many illnesses continue to impose troublingly high health burdens. Recent decades have witnessed the emergence of dramatic new health challenges, including the rise of complex, multi­factorial risk factors such as obesity, antibiotic and drug resistant pathogens, newly emerging infectious diseases that cross the animal­human interface, gaping and rising health disparities, and the emergence of syndemics of interacting illnesses. Although there is widespread recognition of the urgency of identifying and promoting policies that address such health challenges, formulating effective policies is dauntingly complex. While traditional analytical tools can offer much insight into public health gaps, there are significant methodological challenges in using such tools alone to understand the etiology of such disparities, and their implication for policy. These  challenges reflect, among other things, the wide variety of interacting factors playing out over the life of  an individual (often in feedbacks reflecting reciprocal causality), the broad ranges of timescales involved, the multiple levels of causation and effect (e.g., physiological and psychological mechanisms within individuals, surrounding layers of social and institutional context), widespread heterogeneities, the presence of bidirectional, delayed, gene­environment interactions, and discontinuous and time­varying causal interactions between factors within and between scales. Because of the interaction of diverse factors in shaping outcomes, there are also methodological challenges in using traditional approaches unassisted to identify effective, robust and cost­ effective ways to decrease the burden of illness. These interactions operate through many pathways, including immunologic, psycho­social, mental health, nutrition, and through the health care system. Studying such complex linkages purely using one methodology or tool is challenging and, all too often, ineffective. In recent years, a growing number of researchers in public health have addressed similar challenges by complementing traditional techniques with dynamic modeling approaches, such as Agent­Based Modeling, microsimulation, System Dynamics, and dynamic social network analysis. Such studies have shed much light on the interconnections between diverse causal factors, helped inform policy tradeoffs, interpret epidemiological trends, and prioritize data collection. Such models also serve as potent “learning prostheses” to enable us to learn more quickly, deeply, and robustly from the evidence available.

Why Agent-Based Modelling?

Agent­Based Modeling approaches serve as a particularly popular Systems Science technique due to a number of characteristics. Their capacity to capture heterogeneity ­­ both continuous and discrete ­­ at an individual and higher level; such heterogeneity often disproportionately shapes dynamics, and plays a central role in formulating targeted interventions, and is of central importance in understanding health equity concerns, and in formulating interventions that reduce health disparities. Secondly, by characterizing individual trajectories, agent­based models support a life course perspective (permitting, for example, capturing the effects of early life exposures on later life outcomes), as well as interventions depending on such history. Both of these previous considerations further allow for grounding the model using longitudinal data in a way that is not possible using the aggregate perspective of cross­sectional models. Thirdly, such models confer great value for their ability to capture aspects of context. This includes spatial/geographic location of agents ­­ key for understanding, among other considerations, accessibility to resources such as grocery stores offering healthy foods, health care and allied health professional services facilities, walkable built environment, recreational resources, and for investigation of the impact of detailed interventions, such as those involving placement of resources. For many models, capturing context with respect to one or more networks (social networks, sexual networks, familial networks, needle-sharing networks, service networks) is further of great interest, given the role that these networks play in shaping, for example, spread of knowledge, attitudes, beliefs, norms, and innovations.

Reflecting the impacts of such context on choice sets and preferences, Agent­Based models further offer a strong support for capturing aspects of situated individual decision making. Fourthly, the capacity of such models to represent factors at multiple scales ­­ not just at an individual level, but also, for example, family, neighbourhood and region ­­ confers strong support for multi­level health analysis, and understanding the effects of portfolios of interventions that operate at or employ information at such different levels. Finally, the individual­based characterization supported by Agent­Based models often provides key support for securing buy­in from stakeholders who can contribute expertise and evidence speaking to that level of representation, and for whom an individual­level representation will secure additional degrees of buy­in and credibility.

Why Hybrid Modelling?

While each Systems Science technique approach can confer great value when applied in isolation, there are significant, textured and varied tradeoffs that obtain between the methods when applying dynamic modeling approaches to particular problems. Associated considerations – which run the gamut from the capacity to capture longitudinal data, represent situated and localized decision­making, representing aspects network, spatial or multi­level context to communication barriers with stakeholders to performance challenges to human resource and skill sets – mean that the modeling approach best suited to characterizing one component of a system is sometimes poorly suited to others. Attempting to force­fit a single modeling approach onto the modeling effort can lead to diverse adverse outcomes, including – but not limited to – inability to investigate important issues of interest, difficulty in adequately characterizing intervention strategies, and models that are inflexible, or opaque and alienating to stakeholders. Within recent years, advances in modeling technology have opened the door to far more ready creation of hybrid and multi­scale models. In successive problems in diverse domains, such models offer greater capacity to respond to modeler intention, and capable of more precisely addressing research questions involved, as well as more adaptive to changes in modeling priorities and scope in light of model­related learning,  easier to understand and communicate, and more flexible and versatile. In many models, the hybrid approach supports understanding in excess of what would be gained with independent and parallel pursuit of modeling in each distinct modeling approach, and sometimes likely greater than the sum of the insights that could be secured collectively from several single­approach investigations.

Supported by a series of hands­on case studies of hybrid and multi­scale modeling in health and health­specific training examples, this event focuses on practical application of hybrid and multi­scale modeling and the criteria used to select among modeling types. Case studies will support participants in exploring and interacting with diverse health models that combine Agent­Based, System Dynamics and Discrete Event modeling, including ­­ but not limited to ­­ use of agents some of whose states are driven by System Dynamics stocks & flows, System Dynamics models that from which agents are individuated at a certain point in risk progression or where interacting agents drive flow rates, agent­based epidemiological models linked to discrete event models capturing patient care pathways and patient flow, multi­scale models, and case studies combining all three types of modeling. Emphasis will be placed on ensuring that participants can understand adapt the mechanisms in such models to their own circumstances. In addition to hybrid and multi­scale models, we will further examine how multiple modeling methods can be fruitfully combined in the same modeling project to yield ongoing insight.