Welcome to MzeeChakula

MzeeChakula (Swahili for "Elderly Food") is an AI-powered nutrition planning system designed specifically for the elderly population in Uganda. It combines Graph Neural Networks (GNNs) with local knowledge to provide personalized, culturally appropriate meal recommendations.

- :material-robot: **AI-Powered** --- Uses an ensemble of 9 Graph Neural Networks to predict optimal nutrition plans. - :material-food-apple: **Culturally Aware** --- Built on a knowledge graph of local Ugandan foods, cultural practices, and seasonal availability. - :material-account-heart: **Elderly Focused** --- Tailored for specific health conditions like hypertension, diabetes, and frailty common in the elderly. - :material-api: **Modern Stack** --- FastAPI backend, Vue 3 frontend, and Neo4j graph database.

Project Goals

1. Combat Malnutrition: Address the 28% undernourishment rate among Uganda's elderly.
2. Personalization: Move beyond generic advice to specific, actionable meal plans.
3. Accessibility: Provide a voice-enabled interface for ease of use.

Quick Start

Check out the Project Setup guide to get the application running locally.

Architecture Overview

The system consists of three main components:

1. Knowledge Graph: A comprehensive map of foods, nutrients, and conditions.
2. GNN Ensemble: Models that reason over the graph to generate recommendations.
3. Application Layer: A user-friendly interface for interaction.