video1.0<iframe src="https://www.loom.com/embed/34362ef55a4847d78cf2e096010ea564" frameborder="0" width="1728" height="1296" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>12961728Loomhttps://www.loom.com12961728https://cdn.loom.com/sessions/thumbnails/34362ef55a4847d78cf2e096010ea564-00001.gif409.26666666666756Hey there! I'm Jason, a medical student and co-founder of TabMeD. In this Loom, I'm showcasing our patient roleplay simulator, which is a web application that helps medical students diagnose patients with the correct disease using appropriate sympathy and history-taking abilities. I'll be demonstrating a case where I talk to Mrs Chan, a 65-year-old woman with a fever and headache. The system provides a chat bubble interface that instantly pops up, and I have to achieve three learning objectives and score from zero to three. The system is trained on a custom case, and I can ask about medical history, social history, or family history. The conversation is very interactive, and none of the conversations are hard-coded. The back end is hard-coded via Python, and the front end is coded using React and Node.js. The best model we're using here is OpenAI's GPT, but there are another platform local large language models that could be run on local servers. There are huge opportunities in this, and it can be deployed to robots so that students can interact with a real object instead of software on the computer. This can also be further customized into an actual back-end platform where teachers can put in the case, and then there's a student end to log in for student's to use. The teachers can then give feedback and rate the stimulated conversations.