Abstract:

Developing clinical judgment is a core objective in health professions education, yet it remains a complex and often implicit process for students. To address this challenge, we have designed and implemented a pedagogical chatbot aimed at supporting learners in reasoning through clinical situations in a structured and reflective manner. The chatbot was co-developed within a nursing program in collaboration with teachers and aligned with the stages of clinical reasoning.

Integrated into teaching modules on clinical judgment, the chatbot guides students through scenarios requiring hypothesis generation, prioritization of patient problems, and justification of decisions. The conversations are structured around diverse scenarios in which the chatbot takes the role of a patient, prompting reflection, providing feedback, and encouraging metacognitive awareness of reasoning processes. The tool is used during supervised learning sessions with undergraduate students and evaluated through voluntary feedback. The chatbot is also available outside the classroom, allowing students to practice independently whenever they wish.

Initial observations from voluntary student feedback suggest that the chatbot fosters engagement, supports self-directed learning, and makes explicit the reasoning pathways that are often tacit. Teachers reported that it helped students articulate their decision-making more clearly and identify gaps in their clinical analysis. Some usability and technical challenges were reported, underlining the importance of designing conversational agents that balance realism, cognitive load, and pedagogical intent.

This project illustrates how AI-driven conversational tools can enhance critical thinking and reflective practice in health education. Future developments will focus on integrating voice-based interactions and feedback, allowing students to engage in spoken conversations with the simulated patient. We also plan to integrate the chatbot into broader simulation-based learning environments, such as a virtual neighborhood developed for nursing education. In parallel, we aim to test small language models running locally in our institution to reduce the ecological footprint of the tool, ensure data privacy, and address some of the limitations encountered.

Keywords:

Educational technology, Virtual patient, Pedagogical chatbot, Clinical judgment, Artificial intelligence, Health education.