Afyalugha – Swahili AI for Patient Triage
Afyalugha is a Swahili-language AI chatbot helping patients access healthcare information through web, WhatsApp, and SMS in coastal Kenya.
Saved whatsapp number +14155238886 where the AI chatbot will be used by patients
Core Insight: Language remains one of the biggest barriers to healthcare access in East Africa—Afyalugha addresses this directly using localized AI.
Problem
English-only health tools exclude many patients.
Medical AI for African languages is extremely limited.
Coastal Kenya lacks localized digital triage systems.
Objectives
- Deploy Swahili AI chatbot for triage
- Reach 50,000 users in year one
- Reduce language barriers by 40%
- Deploy across clinics in Mombasa, Kilifi, and Kwale
Technical Architecture
- Multi-channel: WhatsApp, SMS, Web
- Swahili NLP pipeline (NER + intent classification)
- Medical knowledge base
- Explainable outputs + clinician review loop
User → NLP → Knowledge Base → Output → Clinician Review
Data Strategy
- 5,000+ Swahili medical terms
- Parallel Swahili-English dataset
- WHO + AMREF validated sources
- Local clinical validation
Timeline
- Months 1–3: Data + partnerships
- Months 4–6: Model development
- Months 7–9: Pilot deployment
- Months 10–12: Scale to 50,000 users
Budget
| Category | Amount |
|---|---|
| Cash | $200,000 |
| Compute | $300,000 |
| Total | $500,000 |
Impact
- 50,000+ users reached
- 85%+ diagnostic support accuracy target
- Improved patient understanding
- Open-source Swahili medical dataset
Extended Vision
Afyalugha is designed not just as a chatbot, but as a scalable digital health infrastructure layer for underserved populations.
- Integration with county health systems
- AI-assisted triage in rural clinics
- Voice-based Swahili interaction (future)
- Regional expansion beyond Kenya
Long-term Goal: Build the largest open Swahili medical language model for Africa.
No comments:
Post a Comment