In northern districts of Zanzibar, herbal medicines are often used to treat disease and protect oneself against ill-wishers. Though common, the use of such medicines can be dangerous—and, in the case of pregnant women, potentially deadly.
Ramla knows that behind closed doors in her village Gamba, Zanzibar, many people rely on the use of herbal medicines like mpopoo or mlakunguru (names of local herbs, like leaves and roots) to treat disease symptoms rather than visiting a health facility. Pregnant women, too, use these medicines in order to protect their pregnancies—but such use often leads to problems like dangerous side effects for the mother, birth defects in the baby, or even miscarriages. Ramla aims to change that and help the women in her village deliver healthy babies. She is a Community Health Volunteer (CHV) with Jamii ni Afya, a digitally-enabled community health program led by the Zanzibar Ministry of Health and supported by D-tree International.

A community midwife displays the herbs she uses to treat her patients (© Kevin Ferguson)
In her work as a CHV, Ramla regularly visits the homes of pregnant women, screening them for danger signs and any risk factors that could indicate the presence of an urgent health issue that must be addressed at a health facility. She also educates pregnant women and caretakers of children under five on vital health information such as the importance of delivering in a health facility, good nutrition during pregnancy and for growing children, and child developmental milestones. For Ramla, however, one of the most important topics to cover when providing education in her village is about the dangers of herbal medicine use. Though this practice had previously been kept within the household, the community health program has allowed CHVs like Ramla to meet people in their homes and warn them of the dangers of such medicines.
With the support of the Wehubit programme of Enabel, D-tree International is using Machine Learning to help CHVs like Ramla to do even more. Machine learning techniques, operationalized by D-tree International, have been employed to create a model to predict risk to pregnant women; then, for those pregnant women who are identified as having a higher risk of perinatal death, the app provides them with additional health care pathways and differentiated services.
For Ramla, this means that each household visit she makes will be different. She knows that each of her pregnant clients has unique circumstances and needs; and now, she can help them meet those needs. As she registers a new pregnant woman, for example, Ramla now receives real-time feedback from her mobile app as she collects information from the woman. Factors like age, living situation, and obstetric history can all be powerful predictors of pregnancy risks—and because the app collects this information about medical history, it also factors in the woman’s use of herbal medicine. If the system identifies a woman as being at higher risk of pregnancy complications, the Jamii ni Afya app can then provide Ramla with the next steps to take to connect the woman with the tailored care that she needs.

A community health volunteer checks in on a mother of young children during a home visit in Zanzibar
(© Mark Leong)
Ramla knows that the challenge remains in her village to spread awareness about the dangers of herbal medicines; this is what makes her role as a CHV so critical in the Jamii ni Afya program. With the help of machine learning technology, however, Ramla is now able to identify high-risk pregnancies and intervene before the most disastrous effects can occur.