Pandemic infection forecasting through compartmental model and learning-based approaches

TL;DR

A compartmental model is developed that includes time-varying infection rates, which are the key parameters that determine the pandemic's evolution, and a hybrid pandemic infection forecasting methodology is presented that integrates compartmental model and learning-based approaches.

Abstract

The emergence and spread of deadly pandemics has repeatedly occurred throughout history, causing widespread infections and loss of life. The rapid spread of pandemics have made governments across the world adopt a range of actions, including non-pharmaceutical measures to contain its impact. However, the dynamic nature of pandemics makes selecting intervention strategies challenging. Hence, the development of suitable monitoring and forecasting tools for tracking infected cases is crucial for designing and implementing effective measures. Motivated by this, we present a hybrid pandemic infection forecasting methodology that integrates compartmental model and learning-based approaches. In particular, we develop a compartmental model that includes time-varying infection rates, which are the key parameters that determine the pandemic's evolution. To identify the time-dependent infection rates, we establish a hybrid methodology that combines the developed compartmental model and tools from optimization and neural networks. Specifically, the proposed methodology estimates the infection rates by fitting the model to available data, regarding the COVID-19 pandemic in Cyprus, and then predicting their future values through either a) extrapolation, or b) feeding them to neural networks. The developed approach exhibits strong accuracy in predicting infections seven days in advance, achieving low average percentage errors both using the extrapolation (9.90%) and neural network (5.04%) approaches.

Figures (7)

• Figure 1: The SIDAREVH model. Schematic representation of the SIDAREVH model used to describe the evolution of the COVID-19 pandemic. The model splits the population into susceptible, vaccinated susceptible, unvaccinated infected detected, vaccinated infected detected, unvaccinated hospitalized, vaccinated hospitalized, recovered and extinct. Model parameters $\zeta$, $\beta_{uu}$, $\beta_{vu}$, $\beta_{vv}$, $\beta_{uv}$, $\gamma_{i}$, $\gamma_{d}$, $\gamma_{\alpha}$, $\gamma_{h}$, $\xi_{i}$, $\xi_{d}$, $\mu_{\alpha}$, $\mu_{h}$ indicate the transition rates between the states, where $\beta_{uu}$, $\beta_{vu}$, $\beta_{vv}$, and $\beta_{uv}$ are considered as time varying while the remaining as constant.
• Figure 2: The model-based estimated infection rate (IR) values associated with the SIDAREVH model. The model-based values of the time-dependent infection rates, using sliding windows of seven and fourteen days: (A) Unvaccinated to Unvaccinated IR: $\beta_{uu}$, (B) Vaccinated to Unvaccinated IR: $\beta_{vu}$, (C) Vaccinated to Vaccinated IR: $\beta_{vv}$, and (D) Unvaccinated to Vaccinated IR: $\beta_{uv}$.
• Figure 3: Model-based estimated versus learning-based predicted infection rate (IR) values associated with the SIDAREVH model.(A) Unvaccinated to Unvaccinated IR: $\beta_{uu}$, (B) Vaccinated to Unvaccinated IR: $\beta_{vu}$, (C) Vaccinated to Vaccinated IR: $\beta_{vv}$, and (D) Unvaccinated to Vaccinated IR: $\beta_{uv}$.
• Figure 4: Prediction results. Schematic representation of the model and learning based prediction results compared to the data. The daily values of prediction refer to the average infected population, considering seven-day windows.
• Figure 5: Flowchart of the steps followed in the proposed methodology.

Theorems & Definitions (1)

• Remark 1