MaternalCare_AI is a lightweight, domain-specific AI assistant that delivers accessible and localized maternal health guidance using ZIP Code data, structured vulnerability indicators, and a curated knowledge base.
- Localized Health Answers: Delivers insights based on ZIP Code Tabulation Area (ZCTA) using CDC's Social Vulnerability Index (SVI)
- Retrieval-Augmented Generation (RAG): Combines vector search with a QA pipeline over maternal health documents
- Gradio Interface: Easy-to-use frontend for real-time maternal care Q&A
- Low-Resource Deployment: Runs in Google Colab for accessible prototyping
| File | Description |
|---|---|
MD_Data_Miner_ITAI2377.ipynb |
Main AI pipeline (Colab-compatible) |
kb_docs.csv |
Embedded maternal health knowledge base |
SVI_2022_US_ZCTA.csv |
Social Vulnerability Index data |
Maternal Health Risk Data Set.csv |
Risk classification data |
Golden_QA_Semantic_Evaluation.csv |
QA evaluation test set |
- Open the notebook in Google Colab.
- Upload the required CSV files listed above.
- Run all cells to load data, build the knowledge base, and launch the Gradio interface.
- Ask maternal health questions like:
- “Where can I get low-cost maternal health support?”
- “What are signs of postpartum depression?”
- Binte Zahra
MIT License — use freely with attribution.