Case Study: Applying Machine Learning to Digital Health Programs
With Sub-Saharan Africa carrying a large proportion of the global burden of HIV, and survival rates attributed to antiretroviral treatment, retention in care of HIV positive patients is a priority for patient health and epidemic control. One of the biggest challenges in reducing transmission of HIV, TB, and other diseases is the treatment of defaulters, or patients who fail to return for treatment. Missing appointments or defaulting from care prevents effective treatment and has the financial consequence of draining regional budgets. Timely identification of patients at high risk of not returning is unlikely in the current system, where operational indicators are frequently collected by hand at the point-of-care. Dimagi, a digital health provider that partners with HIV programs to support clinical care pathways to frontline workers to manage HIV patients and embed service provision in communities, received a grant through the joint Global Partnership for Sustainable Development Data and World Bank Innovation Fund to ask the following question: “Within a digital health system, can we use Machine Learning to prioritize patients most at risk to default, and how would this be designed into the workflow of a frontline worker?”
Read the case study to learn about the approach and its outcomes.