Vaccine hesitancy is a growing issue, as it plays a vital role in reducing mandatory vaccine coverage. Therefore, there has been a lot of interest in understanding how hesitancy is spreading at a high spatio-temporal resolution, since this can allow public health agencies in targeted interventions. A significant challenge for this research is the lack of high-resolution data which indicates hesitancy.
We develop VaxHesSTL framework, which combines a Graph Neural Network (GNN) and a Recurrent Neural Network (RNN) for the prediction of ZIP Code or ZCTA level hesitancy. The GNN uses a ZIP Code level network that captures detailed population level mixing. We train and evaluate VaxHesSTL using a large All Payers Insurance Claims (APCD) dataset for Virginia, consisting of claims from over five million individuals for six years; we use a specific diagnosis code in APCD which indicates refusal of immunization, as an indicator of hesitancy. Experiments on the APCD show that VaxHesSTL outperforms a range of state-of-the-art (SOTA) baselines (which rely only on the historical time series data and do not account for spatial relationships), particularly when the activity-based aggregated contact network is used. Since such a dataset is expensive, we study an active learning approach to optimize the choice of a training set of ZIP Codes that allows the best performance.
Sifat Moon is a Research Scientist for HPC and AI in Health within the Biostatistics and Multiscale Systems Modeling Group in the Computational Sciences and Engineering Division at Oak Ridge National Laboratory (ORNL).
Prior to joining ORNL, she was a Postdoctoral Research Associate in the Network Systems Science and advanced Computing (NSSAC) section at the University of Virginia. She received her Ph.D. in Computer Engineering from Kansas State University in 2021 for her dissertation “Modeling and analysis of stochastic contagion processes over large networks from limited data.”