A SpatioTemporal model for influenza

Stefan Edlund, Michal Bromberg, Gabriel Chodick, Judith Douglas, Daniel Ford, Zalman Kaufman, Justin T Lessler, Rachel Marom, Yossi Mesika, Roni Ram, Varda Shalev, James Kaufman

Research output: Contribution to journalArticle

Abstract

This paper describes an attempt to model seasonal influenza using the SpatioTemporal Epidemiological Modeler (STEM). Ten years of influenza data collected at 49 locations in Israel by the Israeli Center for Disease Control was used to fit the model, and a deterministic SIR(S) compartmental disease model was extended to account for seasonal variation in transmission rates as well as mixing of infected individuals between geographic regions. An adaptive step-size ordinary differential equation solver generated the time series data, and model parameters were fitted to the first few seasonal cycles of the experimental data and compared to subsequent cycles. The model used a sinusoidal seasonal variation of the transmission rate using an exponent parameter, and the best fit exponent was found to be very close to the square root of a sine function (exponent = 0.55), rather than the simple sine function typically used. Other results found that the transmission coefficient was high most of the year and that spatial mixing was high with 90% of people regularly visiting adjacent regions. The study obtained an excellent fit for the first two annual cycles (with less than 3% Root Mean Square error), with rapidly degrading accuracy in subsequent years. Finally, the annual variation in influenza species and strain suggests that a multi-serotype model for seasonal flu should provide better predictive capability. STEM and the mathematical models used in the study are all open source, available at www.eclipse.org/stem. An extension of the model to incorporate multiple serotypes will be studied in the future.

Original languageEnglish (US)
Article numberE9
JournalElectronic Journal of Health Informatics
Volume6
Issue number1
StatePublished - 2011

Fingerprint

Human Influenza
Israel
Centers for Disease Control and Prevention (U.S.)
Theoretical Models
Serogroup

Keywords

  • Compartmental disease models
  • Health
  • Modeling
  • Seasonal influenza
  • Simulation

ASJC Scopus subject areas

  • Health Informatics
  • Public Health, Environmental and Occupational Health

Cite this

Edlund, S., Bromberg, M., Chodick, G., Douglas, J., Ford, D., Kaufman, Z., ... Kaufman, J. (2011). A SpatioTemporal model for influenza. Electronic Journal of Health Informatics, 6(1), [E9].

A SpatioTemporal model for influenza. / Edlund, Stefan; Bromberg, Michal; Chodick, Gabriel; Douglas, Judith; Ford, Daniel; Kaufman, Zalman; Lessler, Justin T; Marom, Rachel; Mesika, Yossi; Ram, Roni; Shalev, Varda; Kaufman, James.

In: Electronic Journal of Health Informatics, Vol. 6, No. 1, E9, 2011.

Research output: Contribution to journalArticle

Edlund, S, Bromberg, M, Chodick, G, Douglas, J, Ford, D, Kaufman, Z, Lessler, JT, Marom, R, Mesika, Y, Ram, R, Shalev, V & Kaufman, J 2011, 'A SpatioTemporal model for influenza', Electronic Journal of Health Informatics, vol. 6, no. 1, E9.
Edlund S, Bromberg M, Chodick G, Douglas J, Ford D, Kaufman Z et al. A SpatioTemporal model for influenza. Electronic Journal of Health Informatics. 2011;6(1). E9.
Edlund, Stefan ; Bromberg, Michal ; Chodick, Gabriel ; Douglas, Judith ; Ford, Daniel ; Kaufman, Zalman ; Lessler, Justin T ; Marom, Rachel ; Mesika, Yossi ; Ram, Roni ; Shalev, Varda ; Kaufman, James. / A SpatioTemporal model for influenza. In: Electronic Journal of Health Informatics. 2011 ; Vol. 6, No. 1.
@article{fc6f46a665084726b1deaee3dde50c20,
title = "A SpatioTemporal model for influenza",
abstract = "This paper describes an attempt to model seasonal influenza using the SpatioTemporal Epidemiological Modeler (STEM). Ten years of influenza data collected at 49 locations in Israel by the Israeli Center for Disease Control was used to fit the model, and a deterministic SIR(S) compartmental disease model was extended to account for seasonal variation in transmission rates as well as mixing of infected individuals between geographic regions. An adaptive step-size ordinary differential equation solver generated the time series data, and model parameters were fitted to the first few seasonal cycles of the experimental data and compared to subsequent cycles. The model used a sinusoidal seasonal variation of the transmission rate using an exponent parameter, and the best fit exponent was found to be very close to the square root of a sine function (exponent = 0.55), rather than the simple sine function typically used. Other results found that the transmission coefficient was high most of the year and that spatial mixing was high with 90{\%} of people regularly visiting adjacent regions. The study obtained an excellent fit for the first two annual cycles (with less than 3{\%} Root Mean Square error), with rapidly degrading accuracy in subsequent years. Finally, the annual variation in influenza species and strain suggests that a multi-serotype model for seasonal flu should provide better predictive capability. STEM and the mathematical models used in the study are all open source, available at www.eclipse.org/stem. An extension of the model to incorporate multiple serotypes will be studied in the future.",
keywords = "Compartmental disease models, Health, Modeling, Seasonal influenza, Simulation",
author = "Stefan Edlund and Michal Bromberg and Gabriel Chodick and Judith Douglas and Daniel Ford and Zalman Kaufman and Lessler, {Justin T} and Rachel Marom and Yossi Mesika and Roni Ram and Varda Shalev and James Kaufman",
year = "2011",
language = "English (US)",
volume = "6",
journal = "Electronic Journal of Health Informatics",
issn = "1446-4381",
publisher = "Health Informatics, Central Queensland University",
number = "1",

}

TY - JOUR

T1 - A SpatioTemporal model for influenza

AU - Edlund, Stefan

AU - Bromberg, Michal

AU - Chodick, Gabriel

AU - Douglas, Judith

AU - Ford, Daniel

AU - Kaufman, Zalman

AU - Lessler, Justin T

AU - Marom, Rachel

AU - Mesika, Yossi

AU - Ram, Roni

AU - Shalev, Varda

AU - Kaufman, James

PY - 2011

Y1 - 2011

N2 - This paper describes an attempt to model seasonal influenza using the SpatioTemporal Epidemiological Modeler (STEM). Ten years of influenza data collected at 49 locations in Israel by the Israeli Center for Disease Control was used to fit the model, and a deterministic SIR(S) compartmental disease model was extended to account for seasonal variation in transmission rates as well as mixing of infected individuals between geographic regions. An adaptive step-size ordinary differential equation solver generated the time series data, and model parameters were fitted to the first few seasonal cycles of the experimental data and compared to subsequent cycles. The model used a sinusoidal seasonal variation of the transmission rate using an exponent parameter, and the best fit exponent was found to be very close to the square root of a sine function (exponent = 0.55), rather than the simple sine function typically used. Other results found that the transmission coefficient was high most of the year and that spatial mixing was high with 90% of people regularly visiting adjacent regions. The study obtained an excellent fit for the first two annual cycles (with less than 3% Root Mean Square error), with rapidly degrading accuracy in subsequent years. Finally, the annual variation in influenza species and strain suggests that a multi-serotype model for seasonal flu should provide better predictive capability. STEM and the mathematical models used in the study are all open source, available at www.eclipse.org/stem. An extension of the model to incorporate multiple serotypes will be studied in the future.

AB - This paper describes an attempt to model seasonal influenza using the SpatioTemporal Epidemiological Modeler (STEM). Ten years of influenza data collected at 49 locations in Israel by the Israeli Center for Disease Control was used to fit the model, and a deterministic SIR(S) compartmental disease model was extended to account for seasonal variation in transmission rates as well as mixing of infected individuals between geographic regions. An adaptive step-size ordinary differential equation solver generated the time series data, and model parameters were fitted to the first few seasonal cycles of the experimental data and compared to subsequent cycles. The model used a sinusoidal seasonal variation of the transmission rate using an exponent parameter, and the best fit exponent was found to be very close to the square root of a sine function (exponent = 0.55), rather than the simple sine function typically used. Other results found that the transmission coefficient was high most of the year and that spatial mixing was high with 90% of people regularly visiting adjacent regions. The study obtained an excellent fit for the first two annual cycles (with less than 3% Root Mean Square error), with rapidly degrading accuracy in subsequent years. Finally, the annual variation in influenza species and strain suggests that a multi-serotype model for seasonal flu should provide better predictive capability. STEM and the mathematical models used in the study are all open source, available at www.eclipse.org/stem. An extension of the model to incorporate multiple serotypes will be studied in the future.

KW - Compartmental disease models

KW - Health

KW - Modeling

KW - Seasonal influenza

KW - Simulation

UR - http://www.scopus.com/inward/record.url?scp=79955456286&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79955456286&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:79955456286

VL - 6

JO - Electronic Journal of Health Informatics

JF - Electronic Journal of Health Informatics

SN - 1446-4381

IS - 1

M1 - E9

ER -