RELATIONAL AGENTS GROUP
Relational Agents are computer agents designed to form long-term, social-emotional relationships with their users. We are investigating the use of these agents in task domains in which human-agent relationships actually improve task outcomes, such as in coaching, counseling, psychotherapy and healthcare.

Computational Models of Health Behavior Change Dialog

The development of ontologies for behavioral medicine is decades behind that for biomedical ontologies, even though behavioral factors such as smoking, diet and physical inactivity are the primary determinants of half of the Healthy People 2010 Leading Health Indicators. In this collaborative effort involving Prof. Timothy Bickmore, Prof. Kenneth Baclawski, Dr. Candace Sidner, Prof. Abby King (Stanford University), and clinical psychologists, we are developing an ontology of concepts and a task model that are relevant to health behavior change, and that can be used to directly drive a health counseling dialog planner. The initial ontology and task model will be within the domain of physical activity promotion, and we will then extend these models to address a second health behavior (increased consumption of fruits and vegetables), using the same behavior change framework, to demonstrate re-usability of the models. The ontology will span all stages of the transtheoretical model, but with a special emphasis on the representation of techniques from Motivational Interviewing for individuals in the precontemplation stage of change. The resulting systems will be evaluated in a randomzied three-arm longitudinal pilot study.

This project is funded by a grant from the NIH National Library of Medicine.

Publications: