SP3.03 Artificially Intelligent Simulated Patients in undergraduate surgical teaching: A single blinded randomized controlled study

Abstract Introduction The acquisition of reasoning skills in surgery can be challenging. Self-directed learning (SDL) outside the clinical arena has become an important tool for learning in undergraduate medical education. There is limited evidence describing the utilization of artificial intelligence (AI) to support medical learning. AI simulated patients may allow for high fidelity interactions focusing on relevant clinical scenarios. We hypothesized that an AI simulated patient (AISP) interaction would be non-inferior to our current educational resources in facilitating learning in early clinical years medical students. Methods This was a single institution randomized controlled study performed at the University of Auckland. The population was a first clinical year undergraduate cohort with the learning theme being cholelithiasis. The intervention group was randomized to interact with an AISP in order to facilitate self-directed learning and reflection. The control group was randomized to use the current educational resources. Both groups then underwent a standardised clinical skills assessment (vCSA). Quantitative and qualitative data was collected. Results There were significantly more distinction grades in the intervention cohort (p=0.037) than in the control cohort. No statistical difference in overall performance between the two groups was found. Qualitative data also suggested widespread satisfaction and enthusiasm for interaction with the AISP. Conclusion AI technology offers novel options in surgical learning. Our study suggests that AISPs are comparable to current resources in developing clinical skills. They may also contribute to an improved academic performance. The AISP cannot replace patient contact but it may have a role in supplementing curricula and assessments.