Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

Integrating Spatiotemporal Features in LSTM for Spatially Informed COVID-19 Hospitalization Forecasting

Created by
  • Haebom

Author

Zhongying Wang, Thoai D. Ngo, Hamidreza Zoraghein, Benjamin Lucas, Morteza Karimzadeh

Outline

This paper presents a novel parallel-stream long-term memory (LSTM) framework for predicting daily hospitalizations per week in the United States. To address the inaccuracy of existing prediction models during the COVID-19 pandemic, we incorporate social hospitalization proximity (SPH), a spatiotemporal feature based on the social connectivity index of Meta. SPH captures the transmission dynamics over space and time by reflecting weekly population movement. Our model captures both short-term and long-term temporal dependencies, and adjusts prediction consistency and error through a multi-horizontal ensemble strategy. The superiority of the proposed model is confirmed by comparative evaluation with COVID-19 prediction hub ensemble models during the delta and Omicron variant spread periods. In particular, during the Omicron variant spread period, the prediction performance is improved by an average of 27, 42, 54, and 69 hospitalizations per week for the 7-day, 14-day, 21-day, and 28-day forecast periods, respectively. We validate the predictive power of SPH through data removal experiments, highlighting the importance of spatiotemporal features in complex epidemiological modeling of infectious disease spread.

Takeaways, Limitations

Takeaways:
Contributes to improving the accuracy of predicting the number of hospitalized patients for infectious diseases such as COVID-19.
Demonstrating the effectiveness of predictive modeling using spatiotemporal features (SPH).
Presenting the utility of parallel stream LSTM framework and multi-horizontal ensemble strategy.
Suggesting the possibility of utilizing external data such as Meta's social connection index.
Limitations:
Further research is needed on the accuracy and generalizability of SPH.
The model's generalization performance to other infectious diseases or regions needs to be verified.
Consideration needs to be given to the computational cost and complexity of the model.
Consideration should be given to data accessibility and bias issues due to meta's data dependency.
👍