skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


Title: Interpreting County-Level COVID-19 Infections using Transformer and Deep Learning Time Series Models
Deep Learning for Time-series plays a key role in AI for healthcare. To predict the progress of infectious disease outbreaks and demonstrate clear population-level impact, more granular analyses are urgently needed that control for important and potentially confounding county-level socioeconomic and health factors. We forecast US county-level COVID-19 infections using the Temporal Fusion Transformer (TFT). We focus on heterogeneous time-series deep learning model prediction while interpreting the complex spatiotemporal features learned from the data. The significance of the work is grounded in a real-world COVID-19 infection prediction with highly non-stationary, finely granular, and heterogeneous data. 1) Our model can capture the detailed daily changes of temporal and spatial model behaviors and achieves better prediction performance compared to other time-series models. 2) We analyzed the attention patterns from TFT to interpret the temporal and spatial patterns learned by the model. 3) We collected around 2.5 years of socioeconomic and health features for 3142 US counties, such as observed cases, and a number of static (age distribution and health disparity) and dynamic features (vaccination, disease spread, transmissible cases, and social distancing). Using the proposed framework, we have shown that our model can learn complex interactions. Interpreting different impacts at the county level would be crucial for understanding the infection process that can help effective public health decision-making.  more » « less
Award ID(s):
2151597
PAR ID:
10499281
Author(s) / Creator(s):
; ; ; ; ;
Publisher / Repository:
IEEE
Date Published:
Journal Name:
IEEE
ISBN:
979-8-3503-4103-4
Page Range / eLocation ID:
266 to 277
Subject(s) / Keyword(s):
Time Series, Deep Learning, Interpretability, Temporal Fusion Transformer, Spatiotemporal, Attention, County-Level COVID-19 prediction
Format(s):
Medium: X
Location:
Chicago, IL, USA
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; ; ; (, IEEE International Conference on Digital Health (ICDH))
    Deep Learning for Time-series plays a key role in AI for healthcare. To predict the progress of infectious disease outbreaks and demonstrate clear population-level impact, more granular analyses are urgently needed that control for important and potentially confounding county-level socioeconomic and health factors. We forecast US county-level COVID-19 infections using the Temporal Fusion Transformer (TFT). We focus on heterogeneous time-series deep learning model prediction while interpreting the complex spatiotemporal features learned from the data. The significance of the work is grounded in a real-world COVID-19 infection prediction with highly non-stationary, finely granular, and heterogeneous data. 1) Our model can capture the detailed daily changes of temporal and spatial model behaviors and achieves better prediction performance compared to other time-series models. 2) We analyzed the attention patterns from TFT to interpret the temporal and spatial patterns learned by the model. 3) We collected around 2.5 years of socioeconomic and health features for 3142 US counties, such as observed cases, and a number of static (age distribution and health disparity) and dynamic features (vaccination, disease spread, transmissible cases, and social distancing). Using the proposed framework, we have shown that our model can learn complex interactions. Interpreting different impacts at the county level would be crucial for understanding the infection process that can help effective public health decision-making. 
    more » « less
  2. ; ; ; ; ; ; (, Journal of Medical Internet Research)
    null (Ed.)
    Background Population mobility is closely associated with COVID-19 transmission, and it could be used as a proximal indicator to predict future outbreaks, which could inform proactive nonpharmaceutical interventions for disease control. South Carolina is one of the US states that reopened early, following which it experienced a sharp increase in COVID-19 cases. Objective The aims of this study are to examine the spatial-temporal relationship between population mobility and COVID-19 outbreaks and use population mobility data to predict daily new cases at both the state and county level in South Carolina. Methods This longitudinal study used disease surveillance data and Twitter-based population mobility data from March 6 to November 11, 2020, in South Carolina and its five counties with the largest number of cumulative confirmed COVID-19 cases. Population mobility was assessed based on the number of Twitter users with a travel distance greater than 0.5 miles. A Poisson count time series model was employed for COVID-19 forecasting. Results Population mobility was positively associated with state-level daily COVID-19 incidence as well as incidence in the top five counties (ie, Charleston, Greenville, Horry, Spartanburg, and Richland). At the state level, the final model with a time window within the last 7 days had the smallest prediction error, and the prediction accuracy was as high as 98.7%, 90.9%, and 81.6% for the next 3, 7, and 14 days, respectively. Among Charleston, Greenville, Horry, Spartanburg, and Richland counties, the best predictive models were established based on their observations in the last 9, 14, 28, 20, and 9 days, respectively. The 14-day prediction accuracy ranged from 60.3%-74.5%. Conclusions Using Twitter-based population mobility data could provide acceptable predictions of COVID-19 daily new cases at both the state and county level in South Carolina. Population mobility measured via social media data could inform proactive measures and resource relocations to curb disease outbreaks and their negative influences. 
    more » « less
  3. ; (, PLOS Medicine)
    Turner, Richard (Ed.)
    Background With the availability of multiple Coronavirus Disease 2019 (COVID-19) vaccines and the predicted shortages in supply for the near future, it is necessary to allocate vaccines in a manner that minimizes severe outcomes, particularly deaths. To date, vaccination strategies in the United States have focused on individual characteristics such as age and occupation. Here, we assess the utility of population-level health and socioeconomic indicators as additional criteria for geographical allocation of vaccines. Methods and findings County-level estimates of 14 indicators associated with COVID-19 mortality were extracted from public data sources. Effect estimates of the individual indicators were calculated with univariate models. Presence of spatial autocorrelation was established using Moran’s I statistic. Spatial simultaneous autoregressive (SAR) models that account for spatial autocorrelation in response and predictors were used to assess (i) the proportion of variance in county-level COVID-19 mortality that can explained by identified health/socioeconomic indicators (R 2 ); and (ii) effect estimates of each predictor. Adjusting for case rates, the selected indicators individually explain 24%–29% of the variability in mortality. Prevalence of chronic kidney disease and proportion of population residing in nursing homes have the highest R 2 . Mortality is estimated to increase by 43 per thousand residents (95% CI: 37–49; p < 0.001) with a 1% increase in the prevalence of chronic kidney disease and by 39 deaths per thousand (95% CI: 34–44; p < 0.001) with 1% increase in population living in nursing homes. SAR models using multiple health/socioeconomic indicators explain 43% of the variability in COVID-19 mortality in US counties, adjusting for case rates. R 2 was found to be not sensitive to the choice of SAR model form. Study limitations include the use of mortality rates that are not age standardized, a spatial adjacency matrix that does not capture human flows among counties, and insufficient accounting for interaction among predictors. Conclusions Significant spatial autocorrelation exists in COVID-19 mortality in the US, and population health/socioeconomic indicators account for a considerable variability in county-level mortality. In the context of vaccine rollout in the US and globally, national and subnational estimates of burden of disease could inform optimal geographical allocation of vaccines. 
    more » « less
  4. ; ; ; ; (, Journal of the Royal Statistical Society Series C: Applied Statistics)
    Abstract Most COVID-19 studies commonly report figures of the overall infection at a state- or county-level. This aggregation tends to miss out on fine details of virus propagation. In this paper, we analyze a high-resolution COVID-19 dataset in Cali, Colombia, that records the precise time and location of every confirmed case. We develop a non-stationary spatio-temporal point process equipped with a neural network-based kernel to capture the heterogeneous correlations among COVID-19 cases. The kernel is carefully crafted to enhance expressiveness while maintaining model interpretability. We also incorporate some exogenous influences imposed by city landmarks. Our approach outperforms the state-of-the-art in forecasting new COVID-19 cases with the capability to offer vital insights into the spatio-temporal interaction between individuals concerning the disease spread in a metropolis. 
    more » « less
  5. ; ; ; ; ; ; ; ; (, Scientific Reports)
    Abstract The objective of this study was to investigate the importance of multiple county-level features in the trajectory of COVID-19. We examined feature importance across 2787 counties in the United States using data-driven machine learning models. Existing mathematical models of disease spread usually focused on the case prediction with different infection rates without incorporating multiple heterogeneous features that could impact the spatial and temporal trajectory of COVID-19. Recognizing this, we trained a data-driven model using 23 features representing six key influencing factors affecting the pandemic spread: social demographics of counties, population activities, mobility within the counties, movement across counties, disease attributes, and social network structure. Also, we categorized counties into multiple groups according to their population densities, and we divided the trajectory of COVID-19 into three stages: the outbreak stage, the social distancing stage, and the reopening stage. The study aimed to answer two research questions: (1) The extent to which the importance of heterogeneous features evolved at different stages; (2) The extent to which the importance of heterogeneous features varied across counties with different characteristics. We fitted a set of random forest models to determine weekly feature importance. The results showed that: (1) Social demographic features, such as gross domestic product, population density, and minority status maintained high-importance features throughout stages of COVID-19 across 2787 studied counties; (2) Within-county mobility features had the highest importance in counties with higher population densities; (3) The feature reflecting the social network structure (Facebook, social connectedness index), had higher importance for counties with higher population densities. The results showed that the data-driven machine learning models could provide important insights to inform policymakers regarding feature importance for counties with various population densities and at different stages of a pandemic life cycle. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email