Med-AI: AI-Powered Preliminary Health Diagnosis Tool

A smart medical assistant that helps users get preliminary health insights by analyzing symptoms through voice and medical image input using AI- fast, accessible, and privacy-friendly.

Comment

Inspiration

Access to timely medical advice remains a challenge for millions, particularly in remote and underserved areas. We wanted to build something that mimics a conversation with a doctor - simple, fast, and empathetic. The idea for MedAI came from observing how many people rely on the internet to self-diagnose, often without context or accuracy. We aimed to build a human-centric, voice-first medical AI that combines image and symptom-based diagnosis into one unified experience, lowering barriers to early medical intervention.

What it does

MedAI Diagnosis is an AI-powered healthcare platform that allows users to:

  • Describe their symptoms using natural voice input
  • Upload medical images such as X-rays, CT scans, or rashes
  • Receive AI-generated preliminary health assessments
  • Hear spoken feedback from the AI for a more natural experience
  • Track their status and generate professional-grade medical reports

MedAI bridges the gap between people and preliminary medical guidance using multimodal AI, helping users take the right next steps in their healthcare journey.

How we built it

  • Frontend: React 18 with TypeScript, Tailwind CSS, and Framer Motion
  • Voice Input: GROQ Whisper API for real-time speech-to-text
  • Image Analysis: Google Gemini Vision API for processing medical images
  • Voice Output: ElevenLabs API for natural text-to-speech responses
  • Backend: Supabase for database, storage, and authentication
  • Security: Row-level security, audit logging, and encrypted storage

Challenges we ran into

  • Ensuring accurate voice input capture in noisy environments
  • Integrating multiple APIs with real-time performance
  • Designing a natural and fluid voice conversation flow
  • Handling multiple file formats and sizes during image upload
  • Implementing secure data practices for handling medical images

Accomplishments that we're proud of

  • Built a fully functional and secure AI-powered diagnosis platform
  • Combined voice and image input for a rich multimodal experience
  • Delivered a responsive and accessible user interface
  • Supported end-to-end patient flow from diagnosis to report generation

What we learned

  • How to design and implement voice-first user interfaces
  • Techniques for combining multimodal AI capabilities effectively
  • The importance of intuitive design in healthcare technology
  • Best practices for managing sensitive health data securely

What's next for Med-AI: AI-Powered Preliminary Health Diagnosis Tool

  • Train domain-specific AI models using medical datasets
  • Support for multilingual voice input and responses
  • Medical professional collaboration for improved accuracy
  • Mobile-first app for wider accessibility
  • Integration with telemedicine services for doctor follow-up

Built With

Share this project:

Updates