AI maternal health
in every language
Maternal Understanding through Language-Aware Multimodal AI in Uganda
Multilingual, multimodal LLMs delivering evidence-based prenatal and postnatal care in Luganda, Runyankore-Rukiga, and Swahili — as text and audio.
AI-generated audio · 0:24
Transcript
"Okugwa kw'omwana mu lubuto kwe kimanyisa nti omwana ali bulamu. Gwanga n'obango bw'omwana…"
What is MULAMU
Maternal Understanding through
Language-Aware Multimodal AI
MULAMU is a research initiative co-created with midwives, community health workers, and clinicians at Mbarara University of Science & Technology (MUST), Uganda.
The Problem
Fewer than 60% of pregnant women in Uganda attend the WHO-recommended four antenatal visits. Language barriers and limited access to trusted health information leave millions of mothers underserved.
Our Approach
We fine-tune multilingual LLMs on community-sourced, clinician-validated maternal health Q&A — delivering evidence-based guidance as text and audio in Luganda, Runyankore-Rukiga, Swahili, and English.
The Impact
Open-source models and freely licensed datasets empower health workers, researchers, and developers to build accessible maternal care tools for low-resource communities across East Africa.
What we build
Language technology
that saves lives
Language Coverage
Serving the languages
spoken at home
In Uganda, fewer than 60% of pregnant women attend the WHO-recommended four antenatal visits. Language barriers, financial constraints, and geographic isolation remain major obstacles. We bring evidence-based maternal health information to communities in their own words.
60%
ANC attendance gap in Uganda
2K+
People reachable in target languages
Luganda
Oluganda
~5M speakers
Runyankore-Rukiga
Runyankore-Rukiga
~3M speakers
Swahili
Kiswahili
~200M speakers
English
English
~1.5B speakers
Research process
From community to model
Community-sourced questions
Real maternal health questions collected from pregnant women, midwives, and community health workers across Uganda and East Africa.
Expert clinical annotation
Clinicians and obstetricians at MUST validate and annotate answers against WHO prenatal care guidelines and regional best practices.
Multilingual model fine-tuning
Base LLMs are fine-tuned on the curated dataset and evaluated across languages for safety and clinical accuracy.
Try it now
Hear maternal health guidance
in your language
Ask a prenatal question in Luganda, Runyankore-Rukiga, Swahili, or English and receive an evidence-based response as text or audio — free and available to all.