Inspiration
What it does
How I built it
Challenges I ran into
Accomplishments that I'm proud of
What I learned
What's next for Maximally AI Shipathon
Inspiration Access to reliable health information is a challenge for millions, especially in rural or underserved areas. Many people rely on unverified internet sources or local hearsay, leading to misinformation and unsafe self-medication. We wanted to create a trustworthy, AI-powered assistant that delivers verified medical information in a simple, accessible way — available to anyone, anytime, in multiple languages.
What it does MediMate AI is an intelligent health information assistant that:
Accepts text or voice input describing symptoms.
Uses OpenAI GPT combined with verified medical databases (MedlinePlus, WHO) to provide accurate, easy-to-understand explanations.
Categorizes cases as urgent or non-urgent to guide users.
Supports multiple languages for accessibility.
⚠️ MediMate AI does not provide prescriptions or replace professional medical advice.
How I built it User Interface – Built with Streamlit for rapid prototyping and clean design.
Natural Language Understanding – Integrated OpenAI GPT API to process symptoms and generate user-friendly explanations.
Medical Data Retrieval – Connected to MedlinePlus API and WHO health datasets for accurate health data.
Urgency Categorization – Implemented simple decision logic to flag severe symptoms (e.g., chest pain, sudden vision loss).
Voice Input (Optional) – Added Google Cloud Speech-to-Text API for accessibility.
Deployment – Hosted the app on Hugging Face Spaces for free public access.
Challenges I ran into Balancing accuracy with simplicity — ensuring the AI output is reliable but still easy for non-medical users to understand.
Finding free, trusted medical APIs — many medical datasets are locked behind paywalls.
Time constraints — building a functional AI + API integration in under 48 hours required cutting non-essential features.
Multi-language support — handling translation without losing medical meaning.
Accomplishments that I'm proud of Built a working AI-powered health assistant in under 48 hours.
Successfully integrated verified medical databases into an AI workflow.
Designed an accessible, mobile-friendly UI that even non-technical users can navigate.
Created a socially impactful project with potential for real-world use.
What I learned API Integration – Learned how to pull structured health data from public APIs and merge it with AI-generated summaries.
Prompt Engineering – Discovered how to guide GPT models to produce safe, medically-aligned outputs.
User-Centric Design – Gained insight into designing for accessibility and clarity.
Rapid Prototyping – Realized the importance of cutting scope to deliver a working MVP within a hackathon window.
What's next for Maximally AI Shipathon Add symptom history tracking so users can monitor trends over time.
Enable offline mode for areas with limited internet connectivity.
Integrate video consultations with certified doctors for urgent cases.
Expand to regional languages using translation APIs like DeepL or Google Translate.
Enhance safety filters to further reduce the chance of misinformation.
Log in or sign up for Devpost to join the conversation.