MediAI: Smart Health Diagnosis was inspired by the urgent need for accessible, instant, and reliable medical guidance, especially for people in underserved regions who often delay care due to uncertainty or cost. The project uses AI to predict diseases based on user-entered symptoms and health data, offering immediate feedback on possible conditions, severity levels, and nearby treatment options tailored to the user's location and budget. Built using NLP for symptom parsing, an XGBoost-based disease classifier, and a rule-based verification system, it combines healthcare intelligence with user-friendly design. The backend was developed with FastAPI and deployed via AWS using Docker containers, while the frontend leverages React and Tailwind for responsiveness and multilingual support. Throughout development, we faced challenges like noisy medical data, model generalization issues, and the need to design for trust and clarity in a sensitive domain. We’re proud to have achieved high accuracy (87% F1-score), integrated affordability-aware provider recommendations, and received strong user feedback from beta testing. In the process, we gained deep insights into working with healthcare data, building explainable models, and optimizing infrastructure costs. Looking ahead, we plan to expand MediAI with voice input, support for regional languages, integration with wearables, and a doctor-ready explainability dashboard to enhance both accessibility and clinical utility.
Built With
- amazon-web-services
- aws-ec2
- docker
- fastapi
- figma
- git
- github
- javascript
- maps
- numpy
- pandas
- postgresql
- python
- react
- render
- rest-apis
- scikit-learn
- tailwind-css
- xgboost
Log in or sign up for Devpost to join the conversation.