MedTrainAI-AI Patient Simulator

• 

  the conversation with a patient

• 

  the start up page

• 

  patient options

• 

  diagnosis results bottom

• 

  diagnosis results

Inspiration

My inspiration came from my passion to medicine and how I thought about all the study tools that exist to study it, but I dont know any major ones that help future doctors practice talking with patients and diagnosing them.

What it does

In this project you have a conversation with a patient of your choice thats been having specific symptoms. There is also an option to generate a completely new patient. Then it rates your diagnosis on how it would most likely be done in real life.

How we built it

I built it using AI studio and with the help of gemini. I gave AI studio a prompt and the idea for my app and then had it make me the app and then just a few other ideas, tweaks and fixes.

Challenges we ran into

At first, the app was too simple and limited, with only a few patient cases. The AI also tended to give away too much information too quickly. I had to refine the prompt and structure multiple times to make the patient responses more realistic and to allow for many different possible diseases and scenarios.

Accomplishments that we're proud of

How advanced the app became and how, AI actually acted like a real patient and there were many different kinds of diagnosis. Also how realistic the analysis is and it doesn't just give unreasonable points.

What we learned

I learned how powerful prompt engineering can be in controlling AI behavior. I also learned how to design an experience that focuses on problem-solving and reasoning rather than simple question-and-answer interaction.

What's next for AI Patient Simulator

I want to improve the voice interaction because it's slower and more laggy than the texting option.

Built With

• aistudio
• gemini