MediScan AI – Project Overview
💡 Inspiration
Healthcare delays are a major issue, especially in rural and busy urban areas where people either ignore symptoms or rely on unreliable internet advice. We wanted to create a solution that gives quick, trustworthy, and accessible health guidance using AI—helping people take action before conditions worsen.
⚙️ What it does
MediScan AI is an intelligent symptom checker that:
Takes user input in the form of symptoms
Uses AI to predict possible diseases
Provides probability-based results
Suggests precautions and next steps
Alerts users if medical attention is needed
It acts as a first-level health assistant, available anytime.
🛠️ How we built it
Collected a symptom-disease dataset for training
Built a machine learning classification model (Random Forest)
Used NLP techniques to understand user input
Developed a simple frontend interface for easy interaction
Connected everything using a backend (Flask)
⚠️ Challenges we ran into
Handling ambiguous symptom inputs from users
Ensuring accurate predictions with limited datasets
Making the system simple yet reliable
Avoiding misleading medical advice
Designing a clean and user-friendly interface in limited time
🏆 Accomplishments that we're proud of
Successfully built a working AI prototype within hackathon time
Achieved good prediction accuracy (~85%)
Created a simple and accessible UI
Addressed a real-world problem with social impact
Delivered a complete end-to-end solution
📚 What we learned
Practical implementation of machine learning models
Importance of clean and structured datasets
Basics of NLP in real-world applications
Team collaboration under time pressure
How to balance innovation with responsibility in healthcare AI
🔮 What's next for MediScan
Add voice input for easier access
Integrate real doctor consultation features
Improve dataset for higher accuracy
Add multi-language support for wider reach
Include image-based diagnosis (e.g., skin conditions)
Launch as a mobile app for real-world
Log in or sign up for Devpost to join the conversation.