Hci Perspectives on Behavior Change Support Systems

In this thesis we have focused on the techniques and principles from the field of human com-puter interaction to design, implement and evaluate the Kristina coaching system. We paidspecial attention to the effects on user perception when feedback was presented in differentformats.Kristina is a multi-device coaching system. It uses data from different input devices tomeasure physical activity and medication intake. Overviews and feedback messages can bereceived on different output. Principles from the field of human computer interaction cancontribute to the design and the evaluation of early stage health behavior change support sys-tems. Without evaluation studies in the traditional clinical sense we were able to investigatethe user perception of the system and whether the system was supporting the users as it wassupposed to do. We have presented the results of a context mapping study and two onlinequestionnaires. These results were input for a requirement analysis for the first prototype ofthe Kristina system. We have presented five user evaluation studies to investigate differencesin user perception between feedback messages presented in plain text or by a virtual human.Results from the different user evaluation studies showed that feedback messages pre-sented by a virtual human did not improve the user perception of the coaching applicationwhen short and frequent messages were received on a smartphone. Feedback messages pre-sented by a virtual human did improve the user perception of the coaching application whenweekly feedback messages were sent to a personal computer or laptop. Results from the logdata of the six week user evaluation study showed trends in the effects on performance andadherence. We found differences between the control group and the text group on the num-ber of uploads of physical activity data and the level of physical activity. Participants fromthe text group showed a higher number of physical activity data uploads compared to par-ticipants from the control group. The level of physical activity of the participants of the textgroup was higher compared to the control group. Glanceablilty, enabling users to understandinformation with low cognitive effort, is important when feedback messages are received ona smartphone. The use of a virtual human does not satisfy the requirements of glanceability.