Date of Award

Spring 5-1-2017

Document Type


Degree Name

Doctor of Science in Information Systems


Business and Information Systems

First Advisor

Jun Liu

Second Advisor

Insu Park

Third Advisor

Viki Johnson


Behavioral Change Support Systems (BCSSs) aim to change users’ behavior and lifestyle. The potential outcomes of these systems make them especially important in areas such as healthcare where these systems could be leveraged to persuade users toward healthy behaviors and then achieve their health goals better. In this regard, health BCSSs have been gaining popularity with the proliferation of wearable devices and recent advances in mobile technologies. Recently, with the promising influence of smartphone applications in supporting healthy lifestyle, researchers have been attracted to explore design principles that can encourage on-going and sustainable use of these persuasive systems. Little research, however, has developed persuasive design principles based on users’ feedback collected from User-Generated Contents (UGC) such as online users reviews. Designing effective persuasive systems must be driven by paramount consideration of what the users need. Analyzing users’ reviews from the actual use has great potential to inform design of these health consumer technologies through providing developers with valuable insights into users’ preferences, experiences and how they interact with these self-care technologies. Therefore, designing a more effective tools that directly meet users’ requirements, wishes and needs. This study extends the existing literature by discovering design principles for health BCSSs based on a systematic analysis of users’ feedback. Specifically, this research demonstrates the use of text mining approach to design health BCSSs. That is, the primary objective of this research project is to identify design principles of BCSSs using usergenerated content (in terms of apps reviews). Mobile diabetes applications are used as the context of this study. A topic modeling technique is employed to first classify the data (user reviews) into different topics. An existing topic modeling algorithm, Latent Dirichlet Allocation algorithm, is used for classification and identification of various topics. These topics are then mapped into design features which are then grouped into 11 design principles. The importance of the design principles is demonstrated through analyzing the relationship between the design principles and user ratings. First, the existence of the design principles in users’ complaints (i.e., 1- or 2-star reviews) is investigated. Second, the relationship between the number of design principles incorporated in apps and user ratings is explored. Overall, the results highlight the necessity of going beyond the techno-centric approach used in current practice and incorporating the social and organizational (i.e. technical support) features into persuasive system design, as well as integrating users with medical devices such as glucose meter and insulin pump and other systems in their usage context.