Designing conversational agents for facilitating patient-physician communication and clinical consultation

Abstract

Pre-consultation chatbots present a unique opportunity to benefit both patients and physicians by facilitating essential information exchange prior to appointments, streamlining the consultation process. However, existing literature on how to design, implement, and evaluate such applications is limited. My thesis addresses this gap through design and evaluation studies with patients and physicians. I use my understanding of physicians' perspectives on synchronous consultations over text messaging to guide the development of a large-language model based pre-consultation chatbot, which I then test with patients in a real-world clinic. My next steps involve developing an interface that physicians can use to review the patient information from the chatbot before the appointment. My thesis contributes to the growing literature on medical large-language model applications in which physician and patient relationships are enhanced, not replaced. It supports a collaboration model where physicians remain responsible for making clinical decisions.