Computational models of chronic diseases: understanding and leveraging complexity

Peer reviewed: 
No, item is not peer reviewed.
Date created: 
2014-04-14
Identifier: 
etd8333
Keywords: 
Agent based modelling
Binge drinking
Insurgency
Microsimulations
Obesity
Social networks
Abstract: 

People struggle to implement interventions that create a significant and sustainable weight loss. This suggests that we need to move beyond viewing obesity as a mere matter of diet and exercise. Instead, we need to accept that the system giving rise to obesity is complex and re-think the ways in which we have approached weight management. This thesis is concerned with the development of microsimulation models that address three features of this complexity at the level of individual food and physical activity behaviours: heterogeneity of individuals and environments, many interactions and loops found between drivers and nonlinearity of the relationships between behaviours. These models were used to explore the dynamics of behaviours, and to support `what-if' scenarios in which we can assess the virtual consequences of possible interventions. In particular, we found that social influences can be as important as environmental influences in changing individuals' weight. However, simulations suggested that intervening in social influences has a larger impact on changing weights by promoting healthy eating than either tax-based or zoning interventions. Our models demonstrated that the structure of communities can be a confounder, as strongly-linked communities reduce the effectiveness of interventions based on social influences. We also highlighted the potential of models to generate social theories. Taking binge drinking as a case study, we generated a set of hypotheses that explains the behaviour of half of the binge drinkers and 4 out of 5 non-binge drinkers. This thesis also made several technical advances. First, we presented an innovative computational framework that supports modelers in expressing how social influences are mediated by the specificities of each individual's context. This is a major improvement over previous microsimulation models that were limited to viewing obesity as contagious. The framework was demonstrated for insurgencies and obesity. Second, we designed and evaluated novel techniques to address heterogeneity when performing a randomized controlled trial for behaviour change. Software was created for both. Finally, we created a system to support practitioners in navigating the maze of factors relevant to their patients, and we used health games to empower patients into the management of their own well-being.

Document type: 
Thesis
Rights: 
Copyright remains with the author. The author granted permission for the file to be printed and for the text to be copied and pasted.
File(s): 
Supervisor(s): 
Diane Finegood
Department: 
Science: Department of Biomedical Physiology and Kinesiology
Thesis type: 
(Thesis) Ph.D.
Statistics: