Zanzibar faces high levels of neonatal mortality as a result of delays or inability to seek care and biological risk factors that go undetected due to lack of contact with providers. Moreover, the health system in general suffers from a lack of resources, especially for Community Health Workers (CHWs).
The Government of Zanzibar is implementing a national digitally-supported community health program to provide essential health, nutrition, and development services to pregnant women and children.
In that framework, machine learning (ML) is an innovative approach that has potential to improve effectiveness and efficiency of service delivery of Maternal Neonatal and Child Health (MNCH).
The project aims to personalize and improve MNCH in Zanzibar by integrating predictive analytics into the national digital community health system using machine learning. This innovation will enable CHWs to pre-identify women with high-risk pregnancies and target prenatal and postnatal services to mitigate risk and improve outcomes.
The solution builds on a highly effective digital system that has been implemented since 2011 in partnership with the Ministry of Health (MOH) in Zanzibar. The system gives the project unique access to comprehensive longitudinal client data that is continuously updated.
The project demonstrates how a machine learning model can successfully be integrated into a mobile decision-support tool that is used by CHWs in a low-income setting to make real-time predictions of clients’ health outcomes. The successful deployment takes into account constraints such as limited internet connectivity and the limited exposure to technology of the CHWs who are using the tool. They have also proven that tailored care pathways can be built into the app, and that CHWs are willing and able to deliver different packages of care to different clients, tailored to each client’s needs.
