Distributed lag inspired machine learning for predicting vaccine-induced changes in COVID-19 hospitalization and intensive care unit admission

Abstract

Distributed lags play important roles in explaining the short-run dynamic and long-run cumulative effects of features on a response variable. Unlike the usual lag length selection, important lags with significant weights are selected in a distributed lag model (DLM). Inspired by the importance of distributed lags, this research focuses on the construction of distributed lag inspired machine learning (DLIML) for predicting vaccine-induced changes in COVID-19 hospitalization and intensive care unit (ICU) admission rates. Importance of a lagged feature in DLM is examined by hypothesis testing and a subset of important features are selected by evaluating an information criterion. Akin to the DLM, we demonstrate the selection of distributed lags in machine learning by evaluating importance scores and objective functions. Finally, we apply the DLIML with supervised learning for forecasting daily changes in COVID-19 hospitalization and ICU admission rates in United Kingdom (UK) and United States of America (USA). A sharp decline in hospitalization and ICU admission rates are observed when around 40% people are vaccinated. For one percent more vaccination, daily changes in hospitalization and ICU admission rates are expected to reduce by 4.05 and 0.74 per million after 14 days in UK, and 5.98 and 1.04 per million after 20 days in USA, respectively. Long-run cumulative effects in the DLM demonstrate that the daily changes in hospitalization and ICU admission rates are expected to jitter around the zero line in a long-run. Application of the DLIML selects fewer lagged features but provides qualitatively better forecasting outcome for data-driven healthcare service planning.

Deep neural network for monitoring the growth of COVID-19 epidemic using meteorological covariates

“…Though information criterion based lag length selection is commonly used in many linear time series models, it is not apparent in machine learning procedures. Thus we follow the distributed lags of time series to form lagged features and follow the distributed lag inspired machine learning for lagged feature selection ( Khan, Hasan, Abedin, Khan, et al, 2022 ).…”

“…Followed by the procedures in Khan, Hasan, Abedin, Khan, et al (2022) and Strobl et al (2008) , we evaluate the importance scores of distributed lags associated with the lagging indices where is the set of integers. Computed importance scores of the selected subset of lagged features when is used in Eq.…”

“…Based on the time-lagged relationship between the predictors (temperature, relative humidity, interaction term of temperature and relative humidity, and government control measures) and the response (COVID-19 reproduction rate), we examine the importance scores of lagged features and utilize the distributed lag inspired supervised learning ( Khan, Hasan, Abedin, Khan, et al, 2022 ) for forecasting the growth of COVID-19 epidemic. We develop a target and verification forecasting scheme for continuous monitoring of the growth of an epidemic towards the targeted control level.…”

Accuracy and Performance of Machine Learning Methodologies: Novel Assessments of Country Pandemic Vulnerability Based on Non-Pandemic Predictors

“…The main purpose of these variables was to convert the Our World in Data COVID-19 timeseries dataset into a supervised learning problem. This enhancement improved and created more robust predictions [47].…”

“…To improve the performance of the RFR and XGBoost models, a distribution lag was applied to the derived case fatality rate variable. Distribution lags play important roles in explaining the short-run dynamic and long-run cumulative effects of features on a response variable [47], [48]. Time lag variables were created for the previous day's, week's, and month's case fatality rate using the shift() method from the Pandas Library in Python.…”