Converse, Connect and Consolidate – The Development of an Artificial Intelligence Chatbot for Health Sciences Education

Abstract

Introduction: Chatbots are intelligent programmes that can conduct conversations with humans primarily through text. Their use has potential in health sciences education as a scalable form of personalised learning, allowing each student to have instantaneous feedback, ask questions in their natural language and progress at their own pace. We developed an artificial intelligence (AI) chatbot as a digital companion for students in the laboratory setting which facilitates learning with anatomy specimens. This study describes the development of our chatbot and presents the stages of its creation from the underlying pedagogy, technological implementation and user acceptability testing. Method: Development of the ‘A.I.natomy Bot’ was undertaken using the AI platform Amazon Web Services (AWS) Lex service and is made available to students in the laboratory via mobile device-scannable QR code which opens a chat interface. The pedagogical basis of the A.I.natomy Bot is the One Minute Preceptor (OMP) which creates short, structured and meaningful interactions between students and teachers. The bot can ask questions, guide, provide personalised feedback, and freely allow students to ask questions about anatomy specimens. User acceptability testing will be undertaken with students at the University of Hong Kong on a single anatomy station with quantitative questionnaire and qualitative evaluation of their chatbot experience. Learning analytics will also be extracted from student-chatbot interactions. Findings: We present the development of the chatbot’s logic structure, interface and our iterative steps in creation of the chat structure centred on head and neck anatomy through educator and student input. We will also present an analysis of the student-chatbot interactions extracted from the recorded conversations which show how a chatbot learns from student-initiated questions. Conclusion: The development of chatbots as a learning adjunct has pedagogically sound roots and has the potential to create an artificially intelligent programme which improves year-on-year via machine learning.