Skip to main content
Jump to: navigation, search

African Swine Fever

Revision as of 04:31, 30 July 2019 by (Talk | contribs) (Adding a description of the scenarios.)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)


back to Downloadable Scenarios

African Swine Fever Scenarios

this page is under construction

This African Swine Fever Scenarios are based on the epidemiological compartment model described by Barongo et al. 2016 in the paper "A Mathematical Model that Simulates Control Options for African Swine Fever Virus (ASFV)".

The STEM projects are separated in two main parts:

  • The first project deals with the re-implementation of the epidemiological compartment model described in Barongo et al. 2016.
  • The second project demonstrate a modified versions of the African Swine Fever Model and there application on a spatial scale.

The purpose was on the one hand to demonstrate the applicability and feasibility to re-implement and expand a model from the scientific literature using STEM.

Re-implementation of the epidemiological compartement model described by Barongo et al. 2016

The epidemiological compartement model by Barongo et al. 2016 was developed to estimate the impact of different intervention scenarios (different vaccinations and bio-interventions). For that purpose they developed an SEICD Model(Susceptible, Exposed, Infected, Carrier and Death) to take advantage of the described disease progression of ASF in the scientific literature.

We re-implemented the model and it's formulas with the "STEM Model Creator" in both scenarios. In the first scenario, we also included the disease dependent population model. described by Barongo et al.

To investigate the outcome of different disease progressions in the population the authors used pert distributed [1] parameters (min, median and max values) collected from literature and own investigations. For each parameter the authors generated 1000 pert distributed values and run 1000 simulations. To solve the ordinary differential equation (ODE) of the SEICD Model the authors used the gillespie algorithm [2].

To re-impement this part of the model, we used the STEM functionality of running Scenarios in Batch Mode [3]. The approach was to create a XML file with the ending ".modifier" which containe the different parameters of the Model and the 1000 pert disrtibuted values. This generated file is used by STEM to run each value pair step by step. The results where logged with the STEM Logger in a CSV. Afterwards we extracted the results programmaticaly from the generated folder structure of STEM and compared the outcome with the results of Barongo et al. To solve the ODE we used the "Finite-difference solver" and the "Runge-Kutta solver" and compared the results. To that time STEM does'nt provide an implementation of the gillespie algorithm [4]. That might cause differences in the outcome of the model.

We also only re-implemented the base model and the bio-intervention described in the paper. We didn't included the vaccination scenarios described by the authors.

The results of the base model were consistent with the results described in the paper. The results of the bio-intervention were not (might be through the different solving algorithms).

The pert distributed values were generated with the "mc2d"-package in R Statistics Software. For the .modifier XML, we first generated the modifier in STEM by using the "Sequence" option in the menu. Afterwards we introduced pogramatically in the .modifier file (stored in the STEM-workspace in the project directory) the "Sequence"/Parameter-values. After a restart all the values were available in STEM. Then we executed the Scenario in "Batch Mode" as Experiment. To increase the simulation speed we switched the option "Pause simulation after each cycle" off in the Simulation Management of STEM (Window-> Preferences-> STEM-> Simulation Management).

Modified version of the epidemiological compartement model described by Barongo et al. 2016

In scenarion 2 of the model we investigated a own aproach based on the paper. We excluded the disease depency of the population model and extended the model by mixing edges ( The purpose was to describe the local behaviour of wild boar groups. Where only a small part of the group is migrating to other groups. The mixing behavoiur of wild boar is more complex then represented in the model and dpendent on different environmental factors. But this was our first simplified approach to investigate the impact of this population characteristic on the spread of the disease, since this wasn't considered by the model of Barongo et al.2016.

For that purpose we generated a square lattice with the same size of the area tested in Barongo's model. Then we splitted the area into smaller squares and distributed the population size as equal groups in this smaller squares. The squares was connected through mixing edges with the same value. Through that there was a constent mixing along the simulation time. We observed a much slower outbreack speed in the population then in the Model by Barongo. Since we hadn't a homogenous Population where each wild boar had contact with another wild boar.

This example was only one approach to showcase the influence of groups of animals and the mixing and investigating the impact. We thought als about the idea of implementing different age groups [5] and different mixing edges to generate a mor relaitic scenario, but it was'nt implemented due to time constraints.

Also we did'nt adjust any of the disease parameters which should be done on a realistic scenario using outbreak data.

Such a migration edge could be also used on a more global scale on bigger population sizes. Mixing Edges or other Migratory edges could be used to simulate inteventions through fencing or even effects like hunting presure (through higher rates in a specific time). This modifications were only discussed but not implemented, since they also required some modifications in STEM itselfs and the lack of data to that time. Random events could be also integrated with different edges.

The presented scenarios are parts of a feasibility study which was performed as masters thesis project.

Install the Scenarios

This archive contains 6 folders that are part of several STEM projects. To run this downloadable scenario, extract every one of these 6 folders into your STEM workspace. Do NOT next the 6 folders in any other folder (they should reside directly in the workspace).

The two project folders named

  • asf_barongo_et_al_validation.asf_barongo_et_al_validation
  • and
  • asf.asf

contain STEM model builder projects. If you observe any errors after extracting the folders to your workspace and importing the project into STEM, you should rebuild these two projects. To do that, in each model builder project, open the "models" subfolder and then double click on the .metamodel files. You want to be in the STEM Designer perspective when you do this. From the Designer perspective click "rebuild" (gear icon) to automatically regenerate the code. See for more detail on the model creator.

Back to the top