Semantic knowledge modeling and evaluation of argument Theory to develop dialogue based patient education systems for chronic disease Self-Management

Background: To improve understanding as well as uptake of health educational material, it should be tailored to informational needs, offer intuitive modes of interaction, and present credible evidence for health claims. Dialogue systems go some way in meeting these requirements, as they emulate interactive and intuitive person-to-person communication. However, most works do not offer a formal model nor modelling process to structure dialogue content, and do not focus on ensuring credibility. Methods: We propose an Extended Model of Argument (EMA) dialogue model and modelling process to support educational dialogue systems. In this dialogue model, computerized arguments directly offer evidence for health claims. EMA further offers “dialogue by design”, where argument structures and interrelations are dynamically leveraged to offer dialogues, instead of relying on predefining discourse flows. We implemented an EMA-based dialogue education system for Juvenile Idiopathic Arthritis (JIA). We performed a qualitative evaluation with JIA health experts involving a Cognitive Walkthrough and Semi-Structured Interview. We applied Directed Content Analysis using categories from the O'Grady framework, and coded sub-themes within those categories using Grounded Theory. Results: We identified 6 sub-themes within the participant feedback pertaining to Quality, Credibility, and Utility. Participants attached strong importance to credibility and found the dialogue system to be a flexible educational tool. Some participants suggested sorting educational items by importance, and presenting only salient knowledge associations to reduce dialogue complexity. Conclusion: Overall, our qualitative evaluation confirmed the following: the ability of EMA to offer credible and appropriate dialogues; and, in general, the utility of dialogue systems to educate JIA patients and their families. In future work, we will revise the system based on evaluation feedback, and perform a more extensive evaluation with patients and caregivers.