Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A new framework for calibrating COVID-19 SEIR models with spatial-/time-varying coefficients using genetic and sliding window algorithms

Huan Zhou, Ralf Schneider

TL;DR

The paper tackles calibrating a spatial-/time-varying SEIR framework (CoSMic) for COVID-19 by introducing a calibration framework that blends overlapping sliding window segmentation with a parallel Elite Genetic Algorithm. The method encodes $\mu$ for 38 regions over 4-week windows, optimizing via RMSE against ICU-demand data to produce near-optimal temporal trajectories of $\mu$ while managing a large, high-dimensional search space. Validation on German regional ICU data shows the approach yields RMSE around $12$ on average and completes within about 1.5 days on HPC hardware, demonstrating feasibility for real-time or near-real-time calibration in changing environments. The framework provides high-level inputs/outputs and enables automated or post-tuning calibration, offering a practical tool for capturing spatial-temporal variations in transmission dynamics due to NPIs and other factors.

Abstract

A susceptible-exposed-infected-removed (SEIR) model assumes spatial-/time-varying coefficients to model the effect of non-pharmaceutical interventions (NPIs) on the regional and temporal distribution of COVID-19 disease epidemics. A significant challenge in using such model is their fast and accurate calibration to observed data from geo-referenced hospitalized data, i.e., efficient estimation of the spatial-/time-varying parameters. In this work, a new calibration framework is proposed towards optimizing the spatial-/time-varying parameters of the SEIR model. We also devise a method for combing the overlapping sliding window technique (OSW) with a genetic algorithm (GA) calibration routine to automatically search the segmented parameter space. Parallelized GA is used to reduce the computational burden. Our framework abstracts the implementation complexity of the method away from the user. It provides high-level APIs for setting up a customized calibration system and consuming the optimized values of parameters. We evaluated the application of our method on the calibration of a spatial age-structured microsimulation model using a single objective function that comprises observed COVID-19-related ICU demand. The results reflect the effectiveness of the proposed method towards estimating the parameters in a changing environment.

A new framework for calibrating COVID-19 SEIR models with spatial-/time-varying coefficients using genetic and sliding window algorithms

TL;DR

The paper tackles calibrating a spatial-/time-varying SEIR framework (CoSMic) for COVID-19 by introducing a calibration framework that blends overlapping sliding window segmentation with a parallel Elite Genetic Algorithm. The method encodes for 38 regions over 4-week windows, optimizing via RMSE against ICU-demand data to produce near-optimal temporal trajectories of while managing a large, high-dimensional search space. Validation on German regional ICU data shows the approach yields RMSE around on average and completes within about 1.5 days on HPC hardware, demonstrating feasibility for real-time or near-real-time calibration in changing environments. The framework provides high-level inputs/outputs and enables automated or post-tuning calibration, offering a practical tool for capturing spatial-temporal variations in transmission dynamics due to NPIs and other factors.

Abstract

A susceptible-exposed-infected-removed (SEIR) model assumes spatial-/time-varying coefficients to model the effect of non-pharmaceutical interventions (NPIs) on the regional and temporal distribution of COVID-19 disease epidemics. A significant challenge in using such model is their fast and accurate calibration to observed data from geo-referenced hospitalized data, i.e., efficient estimation of the spatial-/time-varying parameters. In this work, a new calibration framework is proposed towards optimizing the spatial-/time-varying parameters of the SEIR model. We also devise a method for combing the overlapping sliding window technique (OSW) with a genetic algorithm (GA) calibration routine to automatically search the segmented parameter space. Parallelized GA is used to reduce the computational burden. Our framework abstracts the implementation complexity of the method away from the user. It provides high-level APIs for setting up a customized calibration system and consuming the optimized values of parameters. We evaluated the application of our method on the calibration of a spatial age-structured microsimulation model using a single objective function that comprises observed COVID-19-related ICU demand. The results reflect the effectiveness of the proposed method towards estimating the parameters in a changing environment.
Paper Structure (14 sections, 4 equations, 6 figures, 2 tables, 1 algorithm)

This paper contains 14 sections, 4 equations, 6 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. CoSMic model
  3. Methods
  4. Overlapping sliding window
  5. Genetic algorithm
  6. Representation
  7. The objective function and fitness function
  8. Implementation/Structure of proposed GA heuristic
  9. Proposed framework
  10. High-level structure
  11. System initialization
  12. Calibration algorithm
  13. Results
  14. Conclusion

Figures (6)

  • Figure 1: The schematic of segmentation and overlapping sliding overlapping window
  • Figure 2: Chromosome representation of a candidate solution within certain time window
  • Figure 3: The synthetic workflow of the proposed calibration framework
  • Figure 4: The workflow of system initialization component
  • Figure 5: The workflow of calibration procedure component
  • ...and 1 more figures