Modeling HIV

Designing simulation models for HIV/AIDS

Designing simulation models for HIV/AIDS

Designing simulation models for HIV/AIDS

Human immunodeficiency virus (HIV) has been a major health problem throughout the world for decades. We developed models to understand its dynamics both across hosts, and specifically injection drug users (Eslahchi et al., 2010), and within-hosts. Our model within hosts was used to get a better understanding of how HIV-infected cells behave, and particularly to assess the importance of medication adherence for the effectiveness of treatment. We found that a small lack of adherence could have a proportionally much larger impact on infection as well as trigger negative effects on health sooner (Rana et al., 2015). Consequently, improving adherence can be very beneficial (particularly for those whose adherence is already high), and small issues in adherence should be addressed early on. We are proud of our team’s work on this project, as our 2015 paper was shortlisted for the best paper award at the ACM SIGPADS conference.

Key references:

  • Eslahchi, C., Pezeshk, H., Sadeghi, M., Giabbanelli, P.J., Movahedi, F., Dabbaghian, V. (2010) A probabilistic model for the spread of HIV infection among injection drug users. World Journal of Modelling and Simulation, 6(4):267-273.
  • Rana, E., Giabbanelli, P.J., Balabhadrapathruni, N.H., Li, X., Mago, V.K. (2015) Exploring the relationship between adherence to treatment and viral load through a new discrete simulation model of HIV infectivity. In Proceedings of the 3rd ACM SIGSIM Conference on Principles of Advanced Discrete Simulation.

Key collaborators:

  • Dr. Vijay Mago, Lakehead University, Canada

Complex Networks

Building and Analyzing Network Models

Building and Analyzing Network Models

Building and Analyzing Network Models

Researchers have found that disparate networks, arising in nature or society, share numerous properties. Since the first properties were found in the late 1990s, a considerable body of work has been devoted to designing models of networks expressing given properties that are tunable by the user. Our contributions in this area were twofold. First, we analyzed existing models to better understand their key characteristics. In particular, we proved close-forms formulas for the average path length in deterministic and stochastic networks (Giabbanelli et al, 2010; Giabbanelli et al, 2011). Second, we designed a new model for complex networks, which considerably generalized previous fractal models. This was important because properties demonstrated on this general model would then apply to all sub-models, thus avoiding the need to demonstrate the same aspect many times (Giabbanelli 2011). As this Alcatel-funded project came to an end, we concluded by showing how complex network models were relevant for the design of backbone networks (Giabbanelli 2010).

Key references:

  • Giabbanelli, P.J., Mazauric, D., Perennes, S. (2010) Computing the average path length and a label-based routing in a small-world graph. In Proceedings of the 12th AlgoTel.
  • Giabbanelli, P.J. (2010) Impact of complex network properties on routing in backbone networks. In Proceedings of IEEE GLOBECOM, 389-393.
  • Giabbanelli, P.J., Mazauric, D., Bermond, J-C (2011) On the average path length of deterministic and stochastics recursive networks. Complex Networks, 1-12.
  • Giabbanelli, P.J. (2011) The small-world property in networks growing by active edges. Advances in Complex Systems, 14(6), 853-869.

Key collaborators:

  • Dr. Dorian Mazauric, INRIA, France
  • Dr. Jean-Claude Bermond, CNRS, France

Innovative Pedagogy

Improving Computer Education

Improving Computer Education

In the same ways, as we pursue innovative research, we seek to innovate in university teaching and learning. We do not see learning as the passive acquisition of knowledge. In fact, as options for e-learning emerged, we questioned how computer science lectures could be transformed (Giabbanelli 2009). Our answer was to adopt a student-centered approach, which rewards creativity and emphasizes hands-on projects starting from the very first year (Giabbanelli 2012). The success of interdisciplinary projects is not limited to undergraduate experiences. Our evaluation of a graduate certificate in modeling complex systems showed that matching computer scientists with subject-matter experts could be highly productive if barriers were appropriately addressed (Giabbanelli et al., 2012). While our teaching journey began as e-learning emerged, we continually need to adapt to shifts in education. Most recently, we assessed how computational modeling could be taught given the recent focus on data science (Giabbanelli & Mago 2016).

Key references:

  • Giabbanelli, P.J. (2009) Why having in-person lectures when e-learning and podcasts are available? In Proceedings of the 14th Western Canadian Conference on Computing Education.
  • Giabbanelli, P.J. (2012) Ingredients for student-centered learning in undergraduate computing science courses. In Proceedings of the Seventeenth Western Canadian Conference on Computing Education.
  • Giabbanelli, P.J., Reid, A.A., Dabbaghian, V. (2012) Interdisciplinary teaching and learning in computing science: Three years of experience in the mocssy program. In Proceedings of the Seventeenth Western Canadian Conference on Computing Education.
  • Giabbanelli, P.J., Mago, V.K. (2016) Teaching Computational Modeling in the Data Science Era. Procedia Computer Science, 80, 1968-1977.

New Modeling Tools

Developing innovative modeling techniques

Developing innovative modeling techniques

Developing innovative modeling techniques

Several obstacles have limited our understanding of how socio-environmental factors interact to shape one’s behavior. One obstacle was the difficulty of developing models when the factors are difficult to measure, given uncertainty or inconsistency in data. One of our major contributions has been to develop a methodology that could be used in this setting (Giabbanelli 2013). We further demonstrated that the methodology had several applications. First, it was not limited to systems with high uncertainty: indeed, it could be interfaced with commonly used techniques in order to capture systems with high uncertainty on certain domains but less on others (Giabbanelli et al., 2014). Second, it could be used as part of randomized clinical trials in order to allocate patients to groups given the uncertainty in the factors involved in the clinical outcome (Giabbanelli et al., 2014). Third, it was suitable for application contexts ranging from complex health behaviors to dynamic societal issues unfolding over a large geographical scale (Giabbanelli 2014; Pratt et al. 2012).

Key references:

  • Pratt, S.F., Giabbanelli, P.J., Jackson, P., Mago, V.K. (2012). Rebel with many causes: A computational model of insurgency. In Proceedings of the 2012 IEEE International Conference on Intelligence and Security Informatics (ISI).
  • Giabbanelli, P.J. (2013) A novel framework for complex networks and chronic diseases. Studies in Computational Intelligence, 424, 207-215.
  • Giabbanelli, P.J., Jackson, P., Finegood, D.T. (2014) Modeling the joint effect of social determinants and peers on obesity. Intelligent Systems Reference Library, 52, 145-160.
  • Giabbanelli, P.J., Crutzen, R. (2014) Creating groups with similar expected behavioral response in randomized controlled trials: a fuzzy cognitive map approach. BMC Medical Research Methodology, 14, 130.
  • Giabbanelli, P.J. (2014) Modeling the spatial and social dynamics of insurgency. Security Informatics, 3(1):1-15.

Key collaborator:

  • Dr. Piper Jackson, Simon Fraser University, Canada

Modeling Human Behaviors

Agent-based models in human behaviors

Agent-based models in human behaviors

Agent-based models in human behaviors

When we started in the field of computational methods for obesity research, we knew that an individual received feedback from the environment as well as peers. However, little was known about the relative importance of these two sources of influence. We led the development of the first models that explored how physical activity and food behaviors were shaped by both social and environmental influences (Giabbanelli et al., 2012). Through simulation studies, it became apparent that both sources of influence contributed to a similar order of magnitude to weight change in the population. This finding was particularly important for obesity research at the time, as the nascent modeling community was focusing on the role of peers while downplaying the contribution of the environment. This early model was later improved by detailing the environment, thus allowing to study specific phenomena such as food deserts (Zhang et al., 2014). While these studies have focused on simulating human behavior in overweight and obesity, we have also examined other complex health problems, leading for examples to new models for binge drinking (Giabbanelli & Crutzen, 2013).

Key references:

  • Giabbanelli, P.J., Alimadad, A., Dabbaghian, V., Finegood, D.T. (2012) Modeling the influence of social networks and environment on energy balance and obesity. Journal of Computational Science, 3, 17-27.
  • Giabbanelli, P.J., Crutzen, R. (2013) An agent-based social network model of binge drinking among Dutch adults. Journal of Artificial Societies and Social Simulations, 16(2).
  • Zhang, D., Giabbanelli, P.J., Arah, O.A., Zimmerman, F.J. (2014) Impact of different policies on unhealthy dietary behaviors in an urban adult population: an agent-based simulation model. American Journal of Public Health, 104(7), 1217-1222.

Key collaborators:

  • Dr. Rik Crutzen, University of Maastricht, The Netherlands
  • Dr. Diane T. Finegood, Simon Fraser University, Canada
  • Dr. Donglan Zhang, University of Georgia, USA