A hybrid Neural Network-SEIR model for forecasting intensive care occupancy in Switzerland during COVID-19 epidemics

Abstract

Anticipating intensive care unit (ICU) occupancy is critical in supporting decision makers to impose (or relax) measures that mitigate COVID-19 transmission. Mechanistic approaches such as Susceptible-Infected-Recovered (SIR) models have traditionally been used to achieve this objective. However, formulating such models is challenged by the necessity to formulate equations for plausible causal mechanisms between the intensity of COVID-19 transmission and external epidemic drivers such as temperature, and the stringency of non-pharmaceutical interventions. Here, we combined a neural network model (NN) with a Susceptible-Exposed-Infected-Recovered model (SEIR) in a hybrid model and attempted to increase the prediction accuracy of existing models used to forecast ICU occupancy. Between 1st of October, 2020 - 1st of July, 2021, the hybrid model improved performances of the SEIR model at different geographical levels. At a national level, the hybrid model improved, prediction accuracy (i.e., mean absolute error) by 74%. At the cantonal and hospital levels, the reduction on the forecast’s mean absolute error were 46% and 50%, respectively. Our findings illustrate those predictions from hybrid model can be used to anticipate occupancy in ICU, and support the decision-making for lifesaving actions such as the transfer of patients and dispatching of medical personnel and ventilators.

A multistate model and its standalone tool to predict hospital and ICU occupancy by patients with COVID-19

“…Recently, machine learning and deep learning models have been applied to study the evolution of the pandemic [ [30] , [31] , [32] ]. The use of these techniques is also applied on health management and particularly on hospital occupancy, combining neural network model (NN) with a Susceptible-Exposed-Infected-Recovered model (SEIR) [ 33 ], or comparing Long Short-Term memory (LSTM) network, convolutional neural network (CNN) and their combination [ 34 ].…”

Dynamically adjustable SVEIR(MH) model of multiwave epidemics: Estimating the effects of public health measures against COVID‐19

“…Mathematical models of epidemics can provide simulation of scenarios for understanding both the disease dynamics and the assessment of intervention scenarios. Several models were deployed to monitor and assess the COVID‐19 pandemic 24–29 . Modeling can inform decisions about optimal lockdown level and timing, reopening, and mitigation strategies 30–33 .…”

“…Several models were deployed to monitor and assess the COVID-19 pandemic. [24][25][26][27][28][29] Modeling can inform decisions about optimal lockdown level and timing, reopening, and mitigation strategies. [30][31][32][33] Although mathematical models were used to compare epidemic intervention strategies, 30,[34][35][36] few modeling studies were conducted to compare surveillance designs in real-time and the rapid introduction of changes to COVID-19 mitigation strategies for optimized effects.…”

Design of effective outpatient sentinel surveillance for COVID-19 decision-making: a modeling study

“…Mathematical transmission models can provide insight into the potential benefits of an outpatient sentinel surveillance system by assessing the effects of sampling effort. Mechanistic models have been used to understand disease dynamics, forecast hospital needs, and evaluate intervention scenarios throughout the COVID-19 pandemic (20)(21)(22)(23)(24)(25). Modeling has been used to inform decision-makers about optimal lockdown, reopening and mitigation strategies (26)(27)(28)(29).…”