Neglected Tropical Diseases disproportionately affect populations in Africa and other low- and middle-income countries. Machine learning has potential to improve disease prediction, detection and control, but its use in neglected tropical disease research remains poorly characterized. This scoping review examines the current landscape of machine learning applications for neglected tropical diseases in Africa, identifying trends, gaps and opportunities for future research and implementation. We conducted a scoping review using Joanna Briggs Institute methodology. PubMed and cited references were searched for studies applying machine learning to neglected tropical diseases in Africa. After screening and eligibility assessment, 77 studies were included in the qualitative synthesis. Here we show that most studies focus on schistosomiasis, leishmaniasis, lymphatic filariasis, and soil-transmitted helminthiases. Geo-risk prediction is the most common application while a few studies address disease detection and none focus on drug discovery or intervention optimization. Tree-based and Maximum Entropy models are the most frequently used and commonly reported as best performing. Most studies use small datasets. African institutional leadership, open sharing of data and source code, engagement of programmatic and policy stakeholders, and deployment of models in real-world settings remain limited. Machine learning research for neglected tropical diseases in Africa remains concentrated in a few diseases and applications, with limited translation into practice. Greater investment in local capacity building, equitable collaborations, open data sharing, transfer learning, deployment-focused research, and standardized machine learning workflows could enhance the real-world impact of machine learning for neglected tropical diseases control and elimination. Neglected tropical diseases (NTDs) affect millions of people, especially in Africa and other low- and middle-income countries (LMICs). Machine learning (ML), a form of artificial intelligence, can help prevent and control these diseases. We reviewed published studies to understand how ML is being used for NTDs in Africa. We identified 77 studies, most of which focused on predicting where diseases are likely to occur. Few studies developed tools for diagnosis, treatment planning, or real-world healthcare use. We also found limited involvement of African research institutions and low sharing of data and computer code. Our findings show that ML could improve NTD control, but stronger collaborations, better data sharing and local capacity building, are needed to ensure impact in affected communities. Galiwango et al. present a scoping review of machine learning applications for neglected tropical disease control in Africa. They show that current research is concentrated on geospatial prediction, with limited reproducibility, deployment, African leadership, and real-world implementation despite strong potential for public health impact.
Galiwango et al. (Tue,) studied this question.