Jacaranda Health operates PROMPTS, an artificial intelligence system that responds to pregnancy and newborn care questions submitted via text message across Kenya. The initiative originated unexpectedly when a clinic’s automated appointment reminder system prompted expectant mothers to begin asking health-related questions. As inquiries grew from fewer than 100 daily to approximately 15,000, the organization transitioned from having nurses respond individually to deploying machine learning technology to manage volume while prioritizing urgent cases.
The system operates entirely through SMS, making it accessible to users with basic phones and limited internet connectivity. It functions in English, Swahili, and Sheng, a Nairobi dialect combining both languages, with adaptations for other West African nations including Ghana, Nigeria, and Tanzania. Approximately 7% of messages are automatically flagged as potentially requiring urgent attention and routed directly to nursing staff without delay. In Kenya, where roughly 6,000 women die annually from pregnancy-related complications—often due to delays in accessing care—the service addresses a critical gap in maternal healthcare availability.
Researchers and health policy experts note that PROMPTS represents a shift in public health communication. Traditional messaging flows unidirectionally from health authorities to patients, whereas the AI system centers conversations on questions mothers actually pose, creating what some describe as a fundamental change in patient agency. The platform also generates data identifying healthcare gaps; for instance, administrators observed discrepancies between blood pressure monitoring and breast examination rates reported by users.
Experts caution that such systems function optimally only within regions maintaining functional healthcare infrastructure. Concerns about error handling, patient privacy, and accountability arise when AI processes medical information. Researchers emphasize the importance of understanding how the system responds to mistakes, such as misclassifying urgent messages as routine cases, and whether human oversight adequately catches such failures.
Article Attribution | Read More at Article Source
Article summary produced by Claude AI