MedSmart

HomePage
Upload
Voice
LLM architecture
yes

Inspiration

We were inspired by how we could utilize AI tools in the medical scene. Specifically, when you have questions about anything related to yourself, we often search these questions up on Reddit. But what is on Reddit can't be taken granted since its anonymous. What we aim to achieve is to give reliable medical advice to our users.

What it does

Our product offers users three convenient ways to access reliable medical information: voice, image, and text. Whether users prefer to ask questions through spoken words, upload pictures for analysis, or type in their inquiries, our model delivers high-quality, evidence-based responses using sources from the Harrison medical textbook and sites from our custom Google search API. It integrates reasoning/chain of thought by allowing the model to break down the context and query to give the user the best response possible. Furthermore, it will help draft an email to send to your doctor.

How we built it

For the frontend we utilized React to build it. For the backend we utilized Groq.

Challenges we ran into

We ran into many challenges. One of them includes formatting the AIs response in a way that was readable. We chose to showcase the output in Markdown since regular text was very unreadable at large chunks. By using the ReactMarkdown library, we formatted our markdown properly. Another problem we had was CORS issues between Flask and React, but those just took time debugging. Also internet issues :C

Accomplishments that we're proud of

We're proud of building something that can be put into quick use. Instead of people having to scavenge through the internet for valid information, we condense this process for them. We were really proud of our complex LLM architecture.

What we learned

We learned more about using different technologies like Groq, Chroma, Deepgram, GCP, and Hyperbolic. Specifically, culminating these technologies to make a clean product.

What's next for MedTalk

For next steps, we'd like to use like fetch.ai where we have different agents which can process different sorts of tasks. One thing that would be cool to implement is to automate next steps in the conversation like booking a doctor's appointment