A novel personal digital assistant (PDA)-based system that can automatically detect bouts of moderate or greater walking and deliver health behavior change information to users to increase their levels of physical activity will be developed and evaluated. The system will be an extension of work already performed by the investigators, and will incorporate a validated wireless motion sensor, pattern classification software to identify bouts of walking, and a personified, relational user interface designed to maintain engagement and trust in the tailored behavior change information delivered to users over multiple interactions. The system will be designed to be worn and used continuously by free-living populations and provide users with health behavior change information at the moment it is needed. Users will interact with the PDA via a simulated face-to-face conversation with the animated relational agent, and will conduct a daily progress review and goal-setting session at which time they will schedule specific times they intend to walk on the following day (bouts of 10 minutes or more of moderate or greater intensity). If they complete a scheduled walk, the agent provides immediate social reinforcement. If they fail to initiate a walk at a scheduled time the agent engages them in a problem-solving session in which it attempts to help them overcome the specific obstacle to exercise they are experiencing. In the proposed effort the components of the PDA-based system will be developed, integrated and tested, and a randomized pilot study conducted to: 1) evaluate the efficacy of the PDA-based behavior change intervention for increasing walking; 2) evaluate the effect of timeliness of health behavior change information on walking (time of need vs. retrospective); and 3) compare the efficacy of the personified user interface with that of a text-based interface for delivering health behavior change information on a PDA. The proposed work will make significant contributions to several areas within the science of medical informatics, extending and integrating work in knowledge representation, bio-signal analysis, natural language processing, and consumer health informatics. This research will advance our understanding of the role of time in health behavior change, and result in a model of the temporal relationships among sensor data, user behavior, user goals, and the delivery of health information intended to change behavior.
| Bickmore, Timothy W; Mauer, Daniel; Brown, Thomas (2009) Context Awareness in a Handheld Exercise Agent. Pervasive Mob Comput 5:226-235 |
| Bickmore, Timothy; Gruber, Amanda; Intille, Stephen (2008) Just-in-time automated counseling for physical activity promotion. AMIA Annu Symp Proc :880 |
| Bickmore, Timothy; Gruber, Amanda; Intille, Stephen et al. (2006) A handheld animated advisor for physical activity promotion. AMIA Annu Symp Proc :855 |
