NutriBot
"Your Personal AI Dietitian—Making Healthy Choices Easier"
Created on 1st October 2024
•
NutriBot
"Your Personal AI Dietitian—Making Healthy Choices Easier"
Describe your project
Health Analysis with GenAI leverages Generative AI (GenAI) to provide personalized health analyses of food products tailored to individual users. This solution addresses the challenge of interpreting misleading health labels, enabling consumers to make informed dietary choices.
In-Scope Solution:
The project features OCR Integration, allowing users to upload images of food products to extract vital nutritional information accurately. Through Nutritional Analysis, GenAI algorithms assess the extracted data, identifying beneficial and harmful ingredients while catering to specific dietary restrictions. User Profile Management enables users to input personal information like age and health conditions, allowing tailored health advice based on individual needs. The system also performs Health Claims Verification, assessing product claims against scientific data to ensure accuracy.
Challenges Faced:
The project encountered challenges in data accuracy, particularly with vector databases. Initial attempts using Vertex AI were limiting, leading to a hybrid approach with Tesseract and Gemini for data enhancement.
Future Opportunities:
Future enhancements include integrating LangChain agents for improved analyses and employing SQLAlchemy for efficient user profile management. The project aims to revolutionize dietary engagement and health management for consumers
Challenges we ran into
Challenges I Ran Into:{most important part}
Data Extraction and Quality:
Initial Issue: Storing extracted data in a vector database initially yielded poor accuracy. The integration with Vertex AI did not meet expectations, leading to inconsistent outputs.
Solution: By combining Tesseract for OCR and Gemini for text extraction, I achieved cleaner data processing and improved accuracy in extracting relevant nutritional information.
Temporary Data Storage:
Initial Issue: Managing user profiles and extracted data without cloud storage presented challenges in data accessibility during sessions.
Solution: I utilized global variables to store temporary user data, enabling easy access and modification throughout the application without reliance on external storage.
User Profile Management:
Initial Issue: Collecting comprehensive user profiles while ensuring data privacy and security was a significant challenge.
Solution: I designed a user-friendly interface guiding users through the profile creation process, including validations to ensure accurate health-related information.
Testing and Validation:
Initial Issue: Ensuring the accuracy of health analyses generated by the AI model required extensive validation against reliable nutritional databases.
Solution: I implemented a thorough validation process by cross-referencing AI-generated outputs with established dietary guidelines and expert consultations.
Tracks Applied (1)