The COVID-19 pandemic’s severe impact highlighted the need for accurate and timely hospitalization forecasting to support effective healthcare planning. However, most forecasting models struggled, particularly during variant surges, when they were most needed. This study introduces a novel parallel-stream Long Short-Term Memory (LSTM) framework to forecast daily state-level incident hospitalizations in the United States. Our framework incorporates a spatiotemporal feature, Social Proximity to Hospitalizations (SPH), derived from Meta’s Social Connectedness Index, to improve forecasts. SPH serves as a proxy for interstate population interaction, capturing transmission dynamics across space and time. Our architecture captures both short- and long-term temporal dependencies, and a multi-horizon ensembling strategy balances forecasting consistency and error. An evaluation against the COVID-19 Forecast Hub ensemble models during the Delta and Omicron surges reveals the superiority of our model. On average, our model surpasses the ensemble by 27, 42, 54, and 69 hospitalizations per state at the 7-, 14-, 21-, and 28-day horizons, respectively, during the Omicron surge. Data-ablation experiments confirm SPH’s predictive power, highlighting its effectiveness in enhancing forecasting models. This research not only advances hospitalization forecasting but also underscores the significance of spatiotemporal features, such as SPH, in modeling the complex dynamics of infectious disease spread.
论文
International Journal of Geographical Information Science
PublisherJournal
中文标题
在LSTM中整合时空特征以实现空间感知的COVID-19住院人数预测
English Title
Integrating spatiotemporal features in LSTM for spatially informed COVID-19 hospitalization forecasting
Zhongying Wang Thoai D. Ngo Hamidreza Zoraghein Benjamin Lucas Morteza Karimzadeh a Department of Geography, University of Colorado, Boulder, CO, USAb Heilbrunn Department of Population and Family Health, Columbia University Mailman School of Public Health, New York, NY, USAc Population Council, NY, USAZhongying Wang is a graduate researcher in the Department of Geography at the University of Colorado Boulder, USA. His research focuses on geospatial data science and machine learning for public health. Wang led the data curation, software implementation, methodology design, formal analysis, visualization, and drafting of the manuscript.Thoai D. Ngo is Professor in the Heilbrunn Department of Population and Family Health, Columbia University Mailman School of Public Health, USA. His work focuses on global public health and epidemiological research. Ngo conceived the study, secured funding, validated the analysis, and critically revised the manuscript.Hamidreza Zoraghein is a Research Scientist at the Population Council, New York, USA. He specialises in spatial data analysis and public-health applications. Zoraghein curated data resources and participated in the critical revision of the manuscript.Benjamin Lucas is a Post-doctoral Associate in the Department of Geography at the University of Colorado Boulder, USA. His interests include machine learning for remote-sensing applications. Lucas advised on the study methodology and contributed to the review and editing of the manuscript.Morteza Karimzadeh is Assistant Professor of Geography at the University of Colorado Boulder, USA. His research integrates geospatial data science, remote sensing and deep learning for public-health, social and environmental applications. Karimzadeh originated the study concept, directed the methodology, supervised the research team, oversaw project administration and funding acquisition, and contributed to both the original drafting and critical revision of the manuscript.
发布时间
2025/7/8 15:01:30
来源类型
journal
语言
en
摘要
元数据
来源International Journal of Geographical Information Science
类型论文
抽取状态raw
关键词
PublisherJournal