Systematic mapping of global research on health systems financing and health economics using machine learning
National health systems shape who receives healthcare, how services are delivered, and whether health systems can achieve equitable and sustainable outcomes. Better health can be an engine of development, fostering economic growth by serving as an investment in human capital. There is a need for transparent, organized evidence to support stronger and more efficient health systems, but despite a vast and diverse health financing literature, it remains difficult to synthesize using traditional approaches.
We are producing the first systematic mapping of global research on health systems financing within the broad health economics literature with support from the Wellcome Trust. This study aims to inform future policy, research, and funding priorities by identifying gaps and documenting the empirical evidence on how health resources are raised, pooled, managed, and allocated and the resulting economic and welfare consequences.
Open consultation | Feedback on draft protocolAs part of our commitment to open and transparent research, we are sharing the draft protocol for the systematic mapping and welcome feedback from researchers, practitioners, policymakers, and others working in health systems financing, evidence synthesis, machine learning for research synthesis, global health and health policy. Please share your thoughts and comments by 15 June via this feedback form. |
About the project
This project will produce the first machine learning-assisted evidence map for health systems financing and health economics research. By combining 3ie’s expertise in evidence synthesis networks of leading health financing experts and institutions, the study will systematically identify and classify a vast body of literature, providing a comprehensive overview of the global evidence base and highlighting priorities for future research, policy, and funding.
Following a comprehensive literature search, machine learning models and large language model-based approaches will support study screening, data extraction, and classification. Studies will be coded by relevant themes such as topic and geography, enabling the creation of a dynamic evidence map, open data sets and accompanying visualizations that provide a structured overview of the global evidence base.