Description
This project is a Flask app running GraphRAG for healthcare, made with Vertex AI and Neo4j, to be deployed in a container (Cloud Run or ECS). Initially, a PDF with diseases descriptions is used to enrich the LLM response via RAG. Then, another LLM automatically parses a CSV file with diseases data, generates the Knowledge Graph. After that, an LLM generates a cypher to query the Neo4j KG database and retrieve the possibles diseases, given the patient medical report.
Deployment in Google Cloud Run
Remove user input from app.py and get JSON via Flask.
Add your secrets to Secret Manager
Adapt configurations in config.json
export GCP_PROJECT='your-project'
export GCP_REGION='us-central1'
export AR_REPO='repo-graphrag'
export SERVICE_NAME='flask-app-graphrag'
gcloud artifacts repositories create "$AR_REPO" --location="$GCP_REGION" --repository-format=Docker
gcloud auth configure-docker "$GCP_REGION-docker.pkg.dev"
gcloud builds submit --tag "$GCP_REGION-docker.pkg.dev/$GCP_PROJECT/$AR_REPO/$SERVICE_NAME"
gcloud run deploy "$SERVICE_NAME" \
--port=8080 \
--image="$GCP_REGION-docker.pkg.dev/$GCP_PROJECT/$AR_REPO/$SERVICE_NAME" \
--allow-unauthenticated \
--platform=managed \
--region=$GCP_REGION \
--project=$GCP_PROJECT \
--set-env-vars=GCP_PROJECT=$GCP_PROJECT,GCP_REGION=$GCP_REGION \
--min-instances 1 --max-instances 5 --cpu 1 --memory 2048Mi --concurrency 10