Logo Lanfrica
  • Accueil
  • Atlas
  • Analyses
  • Documentation
  • Sign in

© 2026 Lanfrica. Tous droits réservés. Tous les droits d'auteur des ressources affichées sur le site Web Lanfrica appartiennent aux détenteurs de droits d'auteur d'origine, sauf indication contraire explicite.

johnuwaishe/Health-Insights-Project

Domaine:

healthcarenatural language processing

Type de record:

software
Créateur:
joh
Hôte:
Research-led conversational AI for exploring WHO public-health indicators for Nigeria and supporting evidence-informed decision-making. # Nigeria Health Insights Nigeria Health Insights is a FastAPI application for exploring public-health indicators for Nigeria through a browser-based chat interface and authenticated API. It combines a local FAISS retrieval index with a Hugging Face inference endpoint and can optionally use OpenAI to improve response presentation. > [!IMPORTANT] > This is a research and demonstration project. Its responses may be incomplete or incorrect and are not medical advice, diagnosis, or a substitute for qualified healthcare professionals or official public-health guidance. ## Features - Retrieval-augmented answers grounded in the included health-indicator data - Browser client served by FastAPI - JWT authentication and role-based access for guests, analysts, and administrators - PostgreSQL-backed users, chat history, settings, and analytics - Hugging Face model inference with optional OpenAI response enhancement - Training notebook and exported Colab workflow for the experimental model pipeline ## How it works 1. The application searches the included FAISS index for health-indicator records relevant to a question. 2. Retrieved records are summarized and passed to the configured Hugging Face inference endpoint. 3. If `OPENAI_API_KEY` is configured, the response can be refined for clarity and actionable context. 4. Authenticated conversations and response metadata are stored in PostgreSQL. The repository includes the source CSV files, a prepared JSONL dataset, document chunks, and the FAISS index so the retrieval layer can load at startup. ## Tech stack - Python, FastAPI, and Uvicorn - PostgreSQL and SQLAlchemy - FAISS and Sentence Transformers - Hugging Face Inference - OpenAI API (optional) - Vanilla HTML, CSS, and JavaScript ## Quick start ### Prerequisites - Python 3.10 or later - PostgreSQL - A Hugging Face token with read access - A deployed Hugging Face inference endpoint - An OpenAI API key only if you want the optional response-enhancement step ### Instal …

Visit

github.com

Tasks

question answering

Tags

conversational-aidecision-supportpublic-healthrag