# Fitting a stochastic model of intensive care occupancy to noisy hospitalization time series during the COVID-19 pandemic

**Authors:** Achal Awasthi, Volodymyr M. Minin, Jenny Huang, Daniel Chow, Jason Xu  
**arXiv:** [2203.00229](https://arxiv.org/abs/2203.00229)  
**Date:** 2022-03-01  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2203.00229v4/

## Key points

- The study focuses on estimating ICU occupancy rates during the COVID-19 pandemic.
- Accurate estimation of admission and discharge rates for ICU patients is crucial for decision-making.
- Individual-level hospital data is often not available to modelers in specific areas.
- Researchers aim to develop a method that infers these rates using publicly available daily numbers of occupied hospital and ICU beds.
- The proposed method is based on an immigration-death process that models fluctuations in ICU occupancy.
- Immigration and death rates can depend on covariates such as hospital bed occupancy and daily SARS-CoV-2 test positivity rate.
- Simulation studies show that the proposed method performs well with noisy time series data.
- The statistical framework is applied to real hospitalization data from UCI Health and Orange County, California.
- Hospitalization and positivity rates are found to be crucial factors for modeling ICU stay dynamics.
- Per-patient ICU stay estimates are validated using anonymized patient-level UCI hospital data.

## Summary

The study focuses on the estimation of intensive care unit (ICU) occupancy rates, which is a crucial indicator of healthcare stress during the COVID-19 pandemic. Accurate estimation of admission and discharge rates for ICU patients is essential for reliable decision-making as the pandemic progresses. However, individual-level hospital data is often not available to modelers in specific geographic areas. Therefore, the researchers aim to develop a method that can infer these rates using publicly available daily numbers of occupied hospital and ICU beds. To achieve this, the researchers propose an estimation approach based on an immigration-death process that models fluctuations in ICU occupancy. The framework allows for immigration and death rates to depend on covariates such as hospital bed occupancy and daily SARS-CoV-2 test positivity rate, which may influence changes in ICU operations. Through simulation studies, the researchers demonstrate that their proposed method performs well even with noisy time series data. They then apply their statistical framework to real hospitalization data from the University of California, Irvine (UCI) Health and Orange County, California. By introducing a likelihood-based framework where immigration and death rates can vary with covariates, the researchers find that hospitalization and positivity rates are crucial factors for modeling ICU stay dynamics. They validate their per-patient ICU stay estimates using anonymized patient-level UCI hospital data. Overall, this study provides a flexible and reliable approach for estimating ICU occupancy rates based on publicly available data. The findings emphasize the importance of considering hospitalization and positivity rates when modeling ICU stay dynamics during the COVID-19 pandemic.