PatientSimAI | Devpost

Logo and Cover Slide

💡 Inspiration

Medical students aren't getting the training they want and need to feel confident in interacting with patients. We've spoken to a PhD student at the Stanford School of Medicine who teaches many students currently struggling with this issue, not just here in the USA, but also in various countries around the world (like Chile, Ireland, Singapore, and Brazil) where students lack the same access to patients and practice.

✏️ What it does

PatientSimAI is a web app that simulates patient interactions using AI and GPT4, to train clinical reasoning, improve medical education, and build skills for practical application. We built a platform where professors can input parameters around which AI can craft a conversation, utilize the scenarios we have crafted, and provide a lesson for students.

🛠️ How we built it

Technology Integration: We combined AI with GPT-4 to simulate authentic patient interactions, enhancing the educational modules in our web app.
Development Journey: We began with a basic prototype to test the AI’s conversational capabilities and iteratively refined both the user experience and functionality based on feedback.
Collaboration: We engaged with medical professionals extensively to ensure the scenarios we created were accurate and educational.

🚧 Challenges we ran into

Text-to-Speech Integration: We faced significant challenges integrating effective text-to-speech capabilities, which were essential for realistic patient interactions.
Deployment Setbacks: Deploying the application on a scalable server was tough and caused delays in our testing and feedback phases.
Managing Time: Balancing the complexity of the project within our deadlines was a substantial challenge.
Creating/Deleting Data: Fetching and adding new courses caused a significant amount of failed API requests in the server.

🏆 Accomplishments that we're proud of

Successful Launch: Despite the hurdles, we successfully launched the app, providing a functional and educational platform for medical students.
Integration Achievement: We effectively connected the front end with the back end, allowing for seamless user interactions and efficient data management.## 📚 What we learned

🔮 What's next for PatientSimAI

We have a few roadmap items for PatientSimAI, to monetize it enough to continue providing this service and making it accessible for as many medical students as possible.

Feature Expansion: We plan to introduce more advanced diagnostic tools and more varied patient scenarios.
Pilot Rollout: We are excited about a potential phased pilot rollout at the Stanford School of Medicine, aiming to collect extensive user feedback and further refine the platform. We aim to partner with professors to add the software as an in-class resource, for students to purchase (B2B targeting).
Enhancing Accessibility: We are focusing on user research to ensure the web application is accessible on various devices, incorporating best healthcare practices and inclusive design to maximize accessibility. We want to verify instructors from areas with fewer resources or have an appeal process, to provide free licenses and improve resource accessibility.
Gamification Elements: We plan to implement gamified features to make the learning process more engaging and informative.

👨‍⚕️ Best Health Prize

We focused on the "service accessibility" aspect of the Health track, and created a tool that can be used anywhere with an internet connection. This is particularly important for countries with fewer resources, where students cannot often practice with human patients in controlled yet flexible spaces that are as conducive to learning.

🗣️ Best Interactive Media

PatientSimAI falls under the category of AI education and use speech-to-text to make the interactions feel more challenging (where students have to think on their feet and verbalize their thoughts succinctly) in an almost realistic-yet-gamified environment.

❤️ Final Notes

Endorsement from Marcos, Medical Surgeon: "PatientSimAI will revolutionize medical education by enabling realistic, interactive scenarios that significantly enhance clinical reasoning skills. This AI-driven platform's ability to customize cases to local specifics, from epidemiology to desired learning outcomes, prepares future healthcare professionals to effectively manage the complexities of patient care."

Technology and medical research is improving constantly. It's time medical education improves with it.

Thank you to all the organizers for the opportunity to participate in HackDavis 2024!