EpiMind - Generative AI for Epidemiology

Large Language Model(LLM) for Epidemiology with RAG (Retrieval-Augmented Generation)

1. Retrive data from the CDC website
   • convert html doc to markdown doc (Done)
   • extract the data from the markdown doc
   • store the data in a vector database pgvector of Postgresql to perform vector searches in LlamaIndex
2. RAG model
   • RAG model for epidemiology
   • Named Entity Recognition(NER) for epidemiology
   • Improve search and generation(KNN,ANN,PCA, LDA etc)
   • knowledge graph for epidemiology
   • Agentic-flow model for epidemiology

Meeting Agenda

• Help RG to setup the environment
• Save data into local or container during develop stage
• How to split (chunk) the text? split based on title would always work (eg too long or too short)
• Embedding models for the text

Setup the Environment

docker \
    run \
        --name epimind \
        -d \
        -v <YOUR LOCAL CODE PATH>:/root/ \
        -p 5432:5432 \
        -p 8888:8888 \
        -e POSTGRES_PASSWORD=password \
        ainilaha/epimind:v0.0.3

Try to access: http://localhost:8888

You can interact with docker container with:

docker exec -it epimind /bin/bash

Also, you can build with the image locally:

You can build the image locally with:

docker build -t epimind:lastest . the local building would much smaller.