Inspiration

Accessing healthcare can be confusing and stressful, especially when you're unsure whether something is urgent or even where to go for help. We wanted to build something that not only provides guidance, but also helps users get connected to actual care when needed.

What it does

MedBuddy is an AI-powered medical triage assistant that lets users discuss any medical concern, ranging from minor symptoms to potential emergencies. Based on the conversation, MedBuddy helps users understand the severity of their situation and suggests appropriate steps.

When signs of severe or urgent conditions are detected, MedBuddy automatically locates the nearest clinic or hospital, and can initiate a phone call and reservation on the user's behalf.

How we built it

MedBuddy was built as a full-stack application. The frontend uses HTML/CSS, and OpenStreetMaps to deliver a responsive chat interface and an interactive map. The backend integrates an LLM for conversational triage, real-time WebSocket communication for live transcript updates, and geolocation services to find nearby clinics.

Challenges we ran into

One of the main challenges we ran into was implementing live voice interaction while simultaneously displaying real-time transcripts in the UI.

Another challenge was ensuring accurate user location detection and map info for clinics nearby.

Accomplishments that we're proud of

  • Integrating live voice input with real-time transcript updates
  • Creating a responsive UI with an interactive map
  • Automatically identifying urgent situations and connecting users to nearby care
Share this project:

Updates

posted an update

AI Usage Disclosure; AI (Gemini and Copilot) was used for the following tasks in this project:

  • Brainstorming (and refinement) of backend architecture
  • Debugging of Twillio and ElevenLabs comms
  • Code cleanup
  • Partial code implementation (function-by-function)
  • LLM System prompts refinement
  • General debugging of code
  • Edge case brainstorming (i.e. phone hangups, rejection, etc.)
  • Frontend initial look and feel and help on subsequent tweaks
  • Advice on deployment

Log in or sign up for Devpost to join the conversation.