MedNet

Inspiration

I had a lung infection, but it took some time to diagnose it. I still suffer from the aftereffects of that if it had been detected in an early stage, maybe it wouldn't have affected me this much.

Similarly, millions of people around the world lack access to timely, accurate medical diagnosis, not because treatments don’t exist, but because they don’t have access to trained specialists. This gap is especially visible in rural areas, developing countries, and overburdened healthcare systems.

We wanted to change that.
MedNet was born out of a vision to democratise medical diagnosis using AI, and to connect the entire medical ecosystem — patients, hospitals, and researchers — into a shared platform that could evolve medicine together.

What it does

MedNet is an AI-powered medical diagnosis platform with a social network layer for collaboration. It offers:

• Heart Sound Analysis via electronic stethoscope recording
• Chest X-ray Analysis using computer vision
• Instant diagnostic feedback with risk scoring and confidence levels
• Custom model training for hospitals using their own datasets
• A research portal for accessing datasets, applying for funding, and publishing models

Whether you're a patient in need of quick insights, a hospital looking to improve diagnostics, or a researcher trying to build the next breakthrough, MedNet is built for you.

How we built it

• Frontend: React 18 + TypeScript + Tailwind CSS + Framer Motion for smooth UX
• AI/ML Models: TensorFlow & PyTorch pipelines for signal and image analysis
• Data Base: Superbase
  

Challenges we ran into

• Signal processing of heart sounds was technically challenging
• X-ray image preprocessing and model fine-tuning required experimentation to achieve high accuracy without overfitting
• Balancing the technical complexity with a minimalist, user-friendly interface was a continual design challenge

Accomplishments that we're proud of

• Successfully built a platform that can handles multi-modal diagnostics (audio + image)
• Created a research portal that enables real scientific collaboration and data sharing
• Designed a highly accessible UI with WCAG compliance, screen reader support, and mobile responsiveness

What we learned

• AI in healthcare is about augmentation, not replacement — doctors want tools, not substitutes
• Building for healthcare means thinking about trust, ethics, and explainability
• Human-centered design is as critical as technical accuracy — if it’s not usable, it’s not useful
• Collaboration features like research sharing and custom training pipelines can transform isolated tools into ecosystems

What's next for MedNet

• Adding new diagnostics like ECG analysis and skin imaging
  
• Enabling telemedicine consultations integrated with diagnostic reports
  
• Expanding the AI model marketplace with community-contributed models
  
• Partnering with hospitals and academic institutions worldwide for real-world pilots
  
• Exploring blockchain integration for secure, decentralised health records

Built With

• bolt
• javascript
• netlify
• python
• react
• supabase
• tensorflow