Leontine Alkema: Bayesian Modeling Meets Ministries of Health: Probabilistic Estimation and Forecasting of Health Indicators in Low- and Middle‑Income Countries

Speaker: Leontine Alkema

 

Institution: University of Massachusetts, Amherst

 

 

Abstract: Statistical models are needed for estimating and forecasting of population health indicators in low- and middle-income countries, where data may be sparse or of varying quality. We developed a class of Bayesian transition models for this purpose and discuss how it addresses statistical challenges, including heterogeneous survey quality, use of routine health‑facility data, outlying observations, short‑term shocks, and the need for subnational estimates and short‑term forecasts. We illustrate the approach with two indicators: use of modern contraceptive methods and antenatal care coverage. We also describe how the modeling approach is used in global and country initiatives. We conclude with lessons learned, open challenges, and future directions for the broader field of statistical modeling to support monitoring of demographic and global health indicators. This is joint work with Herb Susmann, Shauna Mooney, and collaborators from the Track20 project and the Countdown to 2030 for Women’s, Children’s and Adolescents’ Health (CD2030) initiative.

 

Bio: Leontine is Professor of Biostatistics at UMass Amherst. She received her PhD in Statistics from the University of Washington in Seattle, focusing on Bayesian modeling and Statistical Demography. She started her PhD with the goal to work at the United Nations but left on an academic track, loving research and teaching, and collaborating with global organizations to improve estimates and projections for demographic and health outcomes.