Inspiration
Navigating over-the-counter (OTC) medicines is a risky game for many individuals with chronic health issues.
- Millions of Canadians suffer from chronic diseases like diabetes, asthma, and hypertension.
- Yet, no personalized system currently advises them whether a medicine is safe based on their unique health profile.
- We wanted to create a digital twin — a system that understands YOU and advises on medicine safety, preventing harmful side effects or interactions.
✨ MediTwin was born — Your AI-powered medicine advisor.
What it does
- Collects personalized health profiles: Chronic conditions, allergies, current medications.
- Analyzes medicines: You simply type a drug name, and MediTwin checks whether it's safe for YOU.
- Fetches official drug data from OpenFDA API (real warnings, side effects, interactions).
- Generates AI-powered explanations using Google Gemini API — in plain, understandable language.
- Provides personalized recommendations (warnings, suggestions, potential risks).
- Gives patients the power of AI-driven personalized health insights before taking any medicine.
How we built it
Frontend (Next.js + React):
- User-friendly forms for health profile input.
- Medicine check interface with real-time interaction.
- LocalStorage used for session persistence.
Backend (Python + Flask):
- REST APIs to handle user profiles and medicine analysis.
- Connected to OpenFDA API to fetch real medicine data.
- Integrated with Gemini AI API for natural language, user-specific explanations.
AI + APIs:
- Gemini API: Personalized explanation generation.
- OpenFDA API: Official FDA drug data like warnings, interactions.
- CORS-enabled Flask backend to bridge frontend and AI.
Challenges we ran into
- ⚙️ Integrating OpenFDA API dynamically — different medicines have different data representations.
- ⚙️ Gemini AI prompt engineering to generate simple but accurate medical explanations.
- ⚙️ CORS issues when connecting frontend (Next.js) and backend (Flask) — resolved with Flask-CORS.
- ⚙️ Handling user-specific context in AI responses to make it feel personalized.
- ⚙️ Data flow & persistence — managing user profiles temporarily without a database (for hackathon speed).
Accomplishments that we're proud of
- ✅ Built a fully functional AI-powered medical twin in just 36 hours.
- ✅ Successfully integrated two major APIs (OpenFDA and Gemini).
- ✅ Generated personalized, user-specific medicine safety reports.
- ✅ Designed a clean and interactive frontend interface for users.
- ✅ Enabled real-time medicine analysis based on user health profiles — never seen before in an OTC analysis app.
What we learned
- 🌐 How to build end-to-end AI-powered apps (frontend + backend + AI APIs).
- 🔑 Importance of accurate data sourcing (FDA's official database).
- 🧠 Prompt engineering for better AI outputs and safer health explanations.
- 💬 How to make AI explanations human-like and understandable.
- 🔗 How to connect multiple technologies smoothly: Next.js, Flask, Gemini, OpenFDA.
What's next for MediTwin
- 🚀 Add Real-time Databases (e.g., Firebase, MongoDB) for persistent user profiles.
- 🤖 Integrate wearable health data (e.g., Fitbit, Apple Health) for even more personalized insights.
- 💊 Suggest safer alternative medications or home remedies using AI.
- 🧑⚕️ Create a Doctor's portal where physicians can validate recommendations.
- 📱 Launch as a mobile-first app for instant access on-the-go.
- 🇨🇦 Partner with Canadian healthcare systems & pharmacies to make OTC medicine safer for everyone.
Log in or sign up for Devpost to join the conversation.