Design of Self-Reflection Support Agents for Medical Decision-Making
Abstract
This study proposes the design of a conversational agent that supports users’ self-reflection in medical decision-making by grounding the dialogue in users’ own values and situational contexts. The agent’s internal structure was designed by referencing emotional attitude estimation of actual chat consultation participants and response characteristics of support providers. Results from dialogue simulations using multiple agents showed that responses attuned to emotions improved users’ emotional states but received low ratings from third parties. This suggests that the role of dialogue agents in counseling support requires a design that flexibly employs responses emphasizing either emotional support or problem clarification, depending on the user’s state and the stage of the dialogue.
Risa Ohara
Master’s Student(M2)
Atsushi Omata
Research Associate
