MAM-AI: An On-Device Medical Retrieval-Augmented Generation System for Nurses and Midwives in Zanzibar
Ren Yi · 2026
Read MAM-AI: An On-Device Medical Retrieval-Augmented Generation System for Nurses and Midwives in Zanzibar by Ren Yi (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-29580.txt.
Introduction
Yi presents MAM-AI, an assistant that answers medical questions for nurse-midwives in Zanzibar and runs entirely on an ordinary Android phone. Maternal and newborn deaths remain high in sub-Saharan Africa, and many nurses lack full midwifery training. Guidelines are long, and connectivity is intermittent, so consulting them during care is hard.
The system embeds each question with a small model and matches it against eighty-seven guideline documents, split into tens of thousands of passages. A compact generator then answers with citations, fully offline. Layered evaluation of the exact deployed setup delivers a surprise: retrieval works well, and the small embedder ranks near cloud systems. The generator is the weak point, since retrieved context fails to improve its answers and can hurt larger models. Yi deploys the more faithful of two candidate generators, and a redesigned prompt cuts its deflection rate sharply.
Catalog record
| CREATOR | Ren Yi |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 119 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.29580