MedGuard

Inspiration

MedGuard was inspired by a common real-world problem: people often consume medicines without fully understanding whether they are safe for them personally.
Medicine safety depends on multiple factors such as age, existing health conditions, food intake, timing, and dosage, yet most platforms only provide generic drug information.

We wanted to build a system that does not just display medicine details, but evaluates medicine safety in context and clearly explains the reasoning behind each decision.

What it does

MedGuard is a medicine safety and decision-support web application.

It allows users to upload a prescription or medicine image, enter their health context, and receive a personalized safety analysis.
The system classifies medicines as safe, risky, or unsafe, along with clear explanations, warnings, and step-by-step intake guidance in both English and Hindi.

MedGuard focuses on explainable, rule-based decisions rather than black-box predictions.

How we built it

MedGuard follows a backend-driven architecture:

• Users authenticate and upload a prescription or medicine image
  
• OCR extracts raw text from the image
  
• AI is used to convert unstructured OCR text into structured medicine data
  
• A rule-based engine evaluates safety using user context such as age, conditions, food state, and timing
  
• The backend returns a structured, explainable safety response that the frontend renders directly
  

This design ensures consistent logic, minimal frontend complexity, and safer decision-making.

Challenges we ran into

• Handling noisy and incomplete OCR output from low-quality images
  
• Designing safety rules that are strict yet flexible
  
• Avoiding over-reliance on AI while still using it meaningfully
  
• Maintaining clear and user-friendly explanations for medical decisions
  
• Carefully positioning MedGuard as a decision-support tool, not a medical diagnosis system
  

Accomplishments that we're proud of

• Building an explainable rule-based engine for medicine safety
  
• Successfully integrating OCR with structured data generation
  
• Delivering personalized, context-aware safety analysis
  
• Providing bilingual instructions (English and Hindi)
  
• Maintaining a clean backend-driven API design
  

What we learned

• Explainability is critical in healthcare-related systems
  
• Rule-based logic can be more trustworthy than black-box models
  
• OCR pipelines require strong error handling and fallbacks
  
• Backend-first design simplifies system reliability and scaling
  
• AI is most effective when used to assist, not replace, core logic
  

What's next for MedGuard

In the future, we plan to:

• Expand the rule engine with verified medical databases
  
• Add support for drug–drug interaction analysis
  
• Improve OCR accuracy for handwritten prescriptions
  
• Introduce advanced analytics while preserving explainability
  
• Enhance accessibility and regional language support
  

MedGuard aims to continue evolving as a trusted, transparent medicine safety assistant.

Built With

• bcrypt
• cloudinary
• express.js
• html
• javascript
• jwt
• llm
• mongodb
• mongoose
• node.js
• nodemon
• ocr
• react
• restful
• tailwind
• tokens