| Issue/Objective |
Medication errors cause over 3 million deaths globally each year - half of all avoidable harm in healthcare. In low-income countries, 4 in 100 people die from unsafe care, yet the clinical decision support infrastructure that catches prescribing errors in high-income settings is largely absent across African health systems. EMR platforms across Nigeria, Kenya, Rwanda, and Ghana are built for billing, not safety - they flag invoices, not dangerous drug combinations. MARMAR was founded to close this gap. Our objective is to deliver AI-powered medication safety as embeddable infrastructure: a FHIR-compliant Clinical Decision Support API that integrates directly into existing EMR systems, and a patient-facing mobile application that enables individuals to check medication safety in real time - in plain language, without medical training. |
| Methodology/Approach |
MARMAR developed a fine-tuned large language model trained on African and global medication contexts, local drug brand names, and regional prescribing patterns. The CDS engine is delivered as a REST API compliant with FHIR R4/R5 standards, enabling EMR vendors to embed medication safety intelligence into their platforms without building it themselves. A companion mobile application extends access directly to patients. The platform was piloted in Nigeria through a partnership with Exheal Systems, targeting 100 hospitals across the country. Implementation began in January 2026 following incorporation of MARMAR Limited in Nigeria and MARMAR Health Inc. in Delaware. The platform has been validated through the Meta Llama Impact Accelerator and accepted into Microsoft for Startups. |
| Results |
The Exheal pilot has reached over 1,000 patients and 200+ healthcare providers across more than 100 hospitals in Nigeria. The mobile application provides real-time medication safety checks accessible to patients without clinical training. Key lessons learned include: (1) EMR integration via API is the fastest and most scalable deployment pathway in low-resource settings - it requires no workflow disruption and no new hardware; (2) patient-facing tools must be designed for low-literacy contexts with voice and conversational interfaces, not form-based inputs; (3) African medication contexts - local generics, regional brand names, and informal drug markets - require purpose-built AI models, not adaptations of Western clinical |
| Discussion/Conclusion |
MARMAR demonstrates that AI-powered medication safety infrastructure can be built, deployed, and scaled from within Africa - by African founders, for African health systems, without dependence on Western clinical databases or proprietary licensing. The platform's dual-channel architecture (B2B EMR API and B2C mobile app) ensures that both institutional and individual access points are covered. As EMR adoption accelerates across the continent and AI costs continue to fall, the conditions for continent-wide medication safety infrastructure have never been more favorable. Reclaiming global health equity requires not just funding and policy, but locally-built technology that serves populations that global health systems have historically overlooked. MARMAR is that technology - and this pilot is proof it works. |
| Presenters and Affiliations |
Ahmad Rufai Yusuf MARMAR |
