The Supply Chain Risk Model (SCRM) is a tool that was designed to evaluate how practices applied under produce safety systems could affect microbial contamination, and, our outcome measure of risk: the risk of a test being positive. Produce supply chains can generally be described as having five main process stages from farm to fork: Primary Raw Material Production, Harvest, Processing, Presentation to Consumer (e.g., retail, restaurant, etc.) and Consumer Handling. Model steps can be added to represent each of these supply chain stages, where any of the following model step types could be selected: increase or reduction, contamination or removal, product test, or risk output test. Each step has a given probability of occurrence.

Our initial test case for this model was a leafy green supply chain contaminated with Shiga-toxin-producing E. coli (STEC). The parameters of this supply chain are described in the paper and details around the parametrized values and distributions can also be found in the Supplemental Materials folder of this GitHub. We described two contamination scenarios (high and low variability) and two interventions added to the baseline system (improved process controls and additional product testing) to give a robust comparison between contamination variability and the tradeoffs between food safety interventions under these contamination scenarios. The base code for the SCRM described in the manuscript can be found here under the SCRM folder.

To facilitate use of our model the SCRM-Lite was developed as an interactive webpage for users to explore the results of the scenarios described in our leafy green supply chain scenario. The code for the SCRM-Lite can be found here under the SCRM-Lite folder.

Below you will find a description of (i) how to run the SCRM-Lite to visualize the results presented in the manuscript and (ii) how to run the base code for the SCRM model and reproduce the results described in the manuscript. For further instructions and how to launch these apps in R, please reference the SCRM Launch Instructions file and select "View raw" to open.

This app was developed and initialized in R 4.3.3 using Shiny. The packages which are required to run this app include: shiny, shinyjs, bslib, shinyWidgets, and ggplot2

Below describes how to run the SCRM-Lite. You will need to access the files in the SCRM-Lite folder of this GitHub.

Open the SCRM-Lite R Project file.

Open the app R file.

Install the required packages (if needed). RStudio can detect required packages and automatically install them if needed.

Click "Run App" (top right corner).

You will be launched into the SCRM-Lite interactive webpage. There are three slider bars in total: Contamination Event Variability, Additional Product Testing, and Process Wash. The default selection is high variability baseline contamination scenario described in the manuscript.

Contamination Event Variability options: Low, High

Additional Product Testing options: None, Some

Process Wash options: Standard, Improved

The following selections in the SCRM-Lite tool match the scenarios in the manuscript:

This app was developed and initialized in R 4.3.3 using Shiny. The packages which are required to run this app include: parallel, lhs, mc2d, bslib, bsicons, shiny, shinyalert, shinyjs, shinyWidgets, DT, and beepr

Below describes how to run the SCRM base code. You will need to access the files in the SCRM folder of this GitHub.

Open the SCRM-Shiny R Project file.

Open the R file titled "app-V1.1.R"

Install the required packages and RTools 4.3 (if needed). RStudio can detect required packages and automatically install them if needed.

Click "Run App" (top right corner).

You will be launched into the SCRM Shiny app.

To load and run the scenarios described in the manuscript go to the "Model" tab. For an explanation of all terms that appear on the "Model" tab, navigate to the "User Guide" tab > Glossary.

Select from the list of Preset Scenarios. Preset Scenarios default to the high variability contamination system. To make steps appear for a preset scenario hit the "Load Scenario" button. The following table describes the preset scenario as it relates to description and any modifications which should need to be made to reproduce the results:

Note that the User Guide tab also explains the baseline scenario terminology for all the steps that are loaded. These explain each step of the baseline scenarios as described by the "Step Name" field in the model.

The default field mass will = 160,000; the default mass unit will = lb; the default iterations will = 100,000. Leave these defaults to reproduce the results of the manuscript.

You can check the "beep" box if you would like your computer to make a sound once the model has completed running all iterations

$\color{#F03C15}{Reminder!!!}$ Double check that you have not inadvertently changed a value for each step. A warning is not tripped if a value is changed after you load a preset scenario. To cross-check the values, navigate to Table 1 of the manuscript for baseline scenarios and manuscript Materials and Methods section Intervention Scenarios for Comparison for intervention scenarios.

Click "Run". Depending on the scenario, you will be notified that there are step(s) present after at risk output test at retail (current model output). This was coded for future use of the model to force the user to check for errors. We parametrized a Consumer Handling stage in our manuscript, but this does not currently affect our decided location of the risk output test. However, these values are still pulled into our baseline. One could simply uncheck the Consumer Handling model step in the SCRM Shiny app to avoid this pop up error, or simply click "Continue."

A small pop-up will appear at the bottom right of the app once the model has began running. This pop up will let you know the model is running and provide a time stamp for the start.

After successfully running the model, a second small pop-up will appear at the bottom right of the app to notify you the model has completed running and a time stamp for which it finished. Your results will appear in 3 results tables at the top of the model tab. The first table summarizes the key results of the model: the overall risk of a positive test, and the number of highest risk lots. The second table shows the number of lots assigned to each risk category*. Each iteration represents a simulated lot. This table also shows the overall risk. The third table shows the number of positive tests at retail for each risk category and the total number of positive tests at retail. *see manuscript for explanation of risk categories, or navigate to the "User Guide" tab of the model > Glossary > Risk Category.

You can save results from a single run by hitting "Download Results" under each table, or by loading them into slots (equivalent to table rows) so that they appear on the Results Comparisons tab. If you are running multiple scenarios, we suggest saving the results to slots after running each scenario. You can save these by selecting your desired slot and clicking the "Save to Slot" button. This will populate your results into two tables on the Results Comparisons tab.

You can view the list of authors in the AUTHORS file.

Corresponding author: Matthew J. Stasiewicz
103 Agricultural Bioprocess Lab
1302 W. Pennsylvania
Urbana, IL, 1361801
USA
+1-217-265-0963
mstasie@illinois.edu

Update pending publication.

This project's code is licensed under the GNU General Public License v3.0 and dataset is licensed the Creative Commons Attribution Share Alike 4.0 International license. Please see the LICENSE.code and LICENSE.dataset files for details.

Funding for the project was made possible by The Center for Produce Safety project 2023CPS08. Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the authors and do not necessarily reflect the view of The Center for Produce Safety. (https://www.centerforproducesafety.org/).