HELIOS third Short-Term Scientific Mission: Geospatial modelling framework optimization for forecasting the global hemoglobinopathy burden

Written by Ms Mikaella Kontopoulou, Cyprus Institute of Neurology and Genetics, Cyprus

The HELIOS-funded Short-Term Scientific Mission at the WorldPop research group significantly advanced the geospatial modelling framework for forecasting the global burden of hemoglobinopathies. Improved data pre-processing strategies were implemented to address data heterogeneity and ensure that the most reliable data points are used to represent each country at the highest resolution possible. In addition, population-based spatial resampling was adopted to allocate data spanning large geographic regions to specific constituent regions of higher resolution.

The geospatial modelling framework itself underwent significant refinement to increase the precision, granularity and geographic coverage of the predictions. New strategies were devised for integrating complex covariates affecting disease distribution such as migration and a pipeline was devised and tested for generating relevant covariate maps from demographic and health survey data. Relevant covariates fell into 2 categories: those that may influence a population’s awareness of the disease which can consequently impact the population’s behavior and/or beliefs towards prevention policies (e.g., religion) and those that may influence a population’s access to health care facilities which can consequently impact the effectiveness of prevention policies (e.g., urbanization). To manage the increased computational demand from the integration of numerous covariates, a less computationally intensive covariate selection methodology was implemented using non-spatial generalized linear models. Moreover, the model was ensured to be adaptable and capable of accommodating data for different hemoglobinopathies and epidemiological parameters.

Planned follow-up activities include continued collaboration with WorldPop researchers to further refine the model and address any emerging challenges. One of the ongoing developments is the proposal of surveillance sites for future research, which aims to identify regions where predictive uncertainty is high but the cost of conducting studies is low due to the availability of healthcare facilities and favorable environmental conditions. Concurrent with this work is the ongoing biocuration and quality appraisal of sources with epidemiological data on hemoglobinopathies and their integration into the IthaMaps database.

A publication outlining this geospatial modelling framework is planned for the hemoglobinopathy and epidemiological parameter with the most data and the broadest geographical coverage. Presently, this framework is being developed for forecasting the prevalence of beta thalassemia and pre-liminary results will be presented in poster format at ASCAT 2024. Interactive and high-resolution predictive maps produced through this modelling framework will be made accessible via IthaMaps, ensuring widespread availability and utility of the research findings. This works lays a robust foundation for effective planning and implementation of health interventions in the regions most affected by hemoglobinopathies while also guiding future research and data collection, in areas where the predictive uncertainty is high and where intervention could be most impactful.