Dr. Abhirup (Abhi) Datta is an Associate Professor in the Department of Biostatistics
at Johns Hopkins Bloomberg School of Public Health.
Dr. Datta’s research focuses on developing statistical methods for geospatial data with applications to air pollution, forestry, ecology, infectious disease modeling, and on Bayesian hierarchical models for multi-source data with applications to global health.
Dr. Datta has been leading highly innovative and impactful research in geospatial AI methods (statistical modeling and machine learning algorithms) for analysis of large and complex environmental data. His work on Nearest Neighbor Gaussian Processes (NNGP) has become one of the most widely used methods for scalable analysis of massive geospatial data. The original NNGP paper led by Dr. Datta is one of the top-5 most cited papers published in the Journal of the American Statistical Association from 2016-2020.
Dr. Datta's recent research delves into studying the foundations of the emerging field of geoAI, which utilizes artificial intelligence (AI) methods for geospatial data analysis. Dr. Datta has pioneered some of the first theoretical studies in this area, demonstrating that the performance of off-the-shelf machine learning algorithms is adversely affected in geospatial analysis due to their lack of explicit geospatial information. Dr. Datta and his team have developed a novel class of geoAI algorithms (geospatial neural networks and random forests) with explicit geospatial awareness. These new methods integrate traditional statistical principles with machine learning, ensuring scalability and the ability to model complex relationships while accounting for geospatial correlations.
Dr. Datta has been developing Bayesian methods to understand disease burden from multi-source data in many low and middle income countries (LMICS). He is currently leading a high impact global health project aimed to improve cause-specific mortality estimates for children and neonates. Dr. Datta has demonstrated that the estimates of mortality in LMICS, often derived from verbal autopsies, are highly biased and has developed Bayesian transfer learning methods for bias-correction of verbal autopsy. He has been recently awarded a grant PI grant by the Bill and Melinda Gates Foundation for broader application of the method to improve cause-specific mortality estimates globally.
