From Epidemiology to Decision Making: A Systems Science Approach to Evaluate Effectiveness of Complex Behavioral Interventions Restricted; Files & ToC

Shao, Yuefan (Summer 2022)

Permanent URL:


Modifiable health behaviors are key to cardiometabolic disease prevention. A significant number of behavioral interventions have been proposed for healthy behavior promotion. However, identifying the most effective type of behavioral intervention for a given population remains challenging due to two main reasons. First, there is a lack of data for behavioral intervention effectiveness evaluation. Second, effectiveness of complex behavior intervention is dependent on multi-level factors, which poses challenges in intervention outcome evaluation. To address these challenges, this dissertation first empirically identified contributing factors associated with patterns of cigarette smoking and physical activity throughout the life-course. In addition, a complex systems modeling approach was used to evaluate effectiveness of different types of behavioral intervention, given target population network characteristics and individual behavior incentive.

Aim 1 characterized trajectories of physical activity and cigarette smoking from early adolescence to adulthood. Using latent class growth mixture model, results showed that there are three sub-groups of individuals sharing similar patterns of physical activity and past 30-day cigarette smoking behavior from early adolescence to adulthood. Age, socio-demographic and early-life psychological factors are important predictors of trajectories for both behaviors.

Aim 2 used social network analysis and regression methods to evaluate the association between social network characteristics and physical activity/cigarette smoking behaviors during adolescence and the adolescence to young adulthood transition. Results suggest that individuals’ health behaviors at younger age are the strongest predictors of health behaviors during young adulthood. In addition, an individual’s social position during adolescence is a predictor for physical activity level during young adulthood but not for cigarette smoking.

Using computational models, Aim 3 showed that when taking into consideration diffusion of interventions within a network, a highly clustered network does not imply the necessity of network-based intervention. Paradoxically, for networks with longer average path length or unknown network structure, incentivizing individuals might be more effective than interventions on popular opinion leader.

Collectively, findings highlighted that both social network and individual-level heterogeneity are key to shaping population level distributions of health behaviors. In addition, researchers need to embrace a systems science lens when evaluating complex behavioral intervention outcomes. 

Table of Contents

This table of contents is under embargo until 19 August 2024

About this Dissertation

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
  • English
Research Field
Committee Chair / Thesis Advisor
Committee Members
Last modified Preview image embargoed

Primary PDF

Supplemental Files