BlindSpot AI — From an Idea to a Working Prototype
I started BlindSpot AI with one simple question:
“What might we be missing?”
I’ve now built a working prototype that analyzes clinical conversations and provides a second layer of intelligent review.
Some of the features I’ve implemented:
- Voice-to-text using faster-whisper
- Symptom and context extraction
- Condition matching using a medical knowledge base
- Blind-spot detection
- Follow-up questions and suggested tests
- Risk flags and consultation scoring
- Communication and bias indicators
- Session memory and logging
- Flask backend and interactive frontend
The goal is not to replace doctors or make final diagnoses.
The goal is to provide an additional layer of scrutiny and help answer:
“What could have been overlooked?”
This project is still evolving. My next focus areas are evidence-based reasoning, better evaluation, uncertainty handling, and building healthcare AI responsibly.
Project: https://github.com/Mohitku-Panigrahi/BlindSpotAI
Log in or sign up for Devpost to join the conversation.