| Societal Challenge | Partner Organization | AI Technology | Expected Impact |
|---|---|---|---|
| Rural healthcare shortages and language barriers limit timely care. | University of Rwanda – CEBE | Multilingual speech AI, computer vision, PPG, and AI-assisted triage. | More accessible symptom screening and referral guidance for rural areas. |
THE CHALLENGE
Rwanda has approximately 1 physician per 10,000-15,000 people, with many rural residents traveling hours for medical care. Community health workers provide essential first-contact care but lack accessible AI-assisted triage tools. Existing digital health solutions often require English literacy or reliable internet access, limiting their usefulness in rural communities.
OUR AI SOLUTION
Ubuzima AI is a Kinyarwanda-first medical triage assistant combining multilingual voice AI, medical computer vision, smartphone vital signs, and AI-guided triage. The architecture prioritizes lightweight, offline-capable mobile triage with more computationally intensive models ran on local GPU infrastructure.
PROGRESS SINCE PHASE 1
| Phase 1 | Phase 2 Progress |
|---|---|
| Established University of Rwanda partnership, defined Kinyarwanda first triage workflow. | Implemented 3 Kinyarwanda ASR models and Kinyarwanda TTS using local GPU inference. |
| Defined clinical workflow, optical vital-sign sensing, and referral requirements. | Implemented 12 medical image analysis models with explanatory outputs. |
WHY IT MATTERS
PEOPLE
Rural Rwandan patients, especially low-literacy and Kinyarwanda-speaking populations, and CHWs who provide frontline medical care.
IMPACT
Improve access to preliminary health screening and referral guidance. Target 100,000+ annual AI-assisted screenings, ≥15% fewer avoidable clinic trips, and ≥95% sensitivity for urgent referrals.
WORKING WITH PARTNERS
The Regional Centre of Excellence in Biomedical Engineering and eHealth
CEBE provides local healthcare context, identified needs through rural district engagement, and a pathway for CHW recruitment and field validation.
PATH TO SCALE
Research → Prototype → Pilot → Deployment → Regional / National Scale
PIntegrate independently evaluated AI models into a succinct workflow, conduct CHW-guided usability testing, and optimize for low-cost devices. Following validation, the platform could expand into other underserved East African communities..
PHASE 2 GOALS
- Engagement: collaborate with healthcare workers to evaluate accessibility and needs.
- Deployment: Pilot Ubuzima AI with CHWs.
- Clinical Validation: Compare AI triage recommendations against clinician assessments.
- Impact: Quantify screening accessibility, referral accuracy, and user satisfaction.
THE TEAM
- Dennis Kountouris
- Yunjie Tong
- Weldon School of Biomedical Engineering