Abattoir Inspect

AbattoirInspect is an interactive data visualisation and surveillance platform developed by SRUC to support the monitoring of animal health, welfare, and public health using abattoir meat inspection data. The tool enables users to explore trends, benchmark performance, and identify potential health and welfare issues through intuitive visualisations and reporting functions. AbattoirInspect Teaser

The development of AbattoirInspect was funded through the European Union Horizon 2020 DECIDE project (Data-driven Control and Prioritisation of Non-EU-Regulated Diseases). Further information about the DECIDE project and its outputs is available at DECIDE

AbattoirInspect is has been developed using open-source software (R and R Shiny). The source code, documentation, and related project outputs are available through the DECIDE project's Zenodo repository. To access these materials, visit Zenodo and search for DECIDE project AbattoirInspect

License Agreement

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. By using this software, you agree to abide by the terms and conditions of the GPL.

How to Use AbattoirInspect

The application is organised into multiple tabs, each providing a different view of carcase rejection and animal health information, from high-level summaries to detailed farm-level and time-series analyses. This brief explainer video guides new users through the main features of the tool. Abattoir Explainer

Upload Data

The Upload Data tab allows users to either continue using sample (mock) data or upload their own dataset for analysis. This section also provides guidance on the required data structure and formatting to ensure successful upload and compatibility with the dashboard. Be sure to check the required data format, adapt your dataset accordingly, and save it as a comma separated values (csv) file. In addition to the recommendations left in the Upload tab, here is the metadata explanation for the format required. Meta data explanation. We recommend that users explore the tool using the mock data which is already pre-loaded before adapting and uploading their own datasets.

How to upload FSS open data

The AbattoirInspect app has been adapted to load Food Standards Scotland (FSS) open data. If you are wondering how to do this, a step-by-step document on how to do so has been prepared and can be found here How to use FSS data.

National Overview

The National Overview tab provides a high-level summary of carcase rejections data across all farms and abattoirs included in the data uploaded. This could be considered a Regional or Group overview depending on the coverage of the data uploaded. The dashboard is divided into three sections, presenting totals as well as breakdowns for the last quarter, month, and week. These summaries cover full, partial, and total rejections.

Annual trends are visualised using bar plots, enabling users to quickly identify patterns in condemnation rates over time. Additional figures rank the most common causes of full and partial condemnations, highlighting the conditions associated with each category and their relative frequency. This allows users to understand the main meat inspection outcomes at a national or regional level, depending on the coverage of the data uploaded.

Customisable selection features to facilitate date selection (e.g. custom dates; year/quarter; or year/month) and display values (absolute vs relative (percentage)) give users full control of the statistics they want to visualise.

My Farm(s)

The My Farm(s) tab provides benchmarking functionality for selected farms. Users can select one or more farms to analyse the most common full and partial rejections for those farms for the desired timeframe. In the same time, a benchmarking plot compares the performance of the selected farms across the top five most common rejection conditions, allowing users to assess how individual farms perform relative to others in the dataset and relative to the national or group average. This comparative view helps identify performance trends, outliers, and potential areas for improvement.

Time-series and Statistics

The Time-series and Statistics tab enables in-depth temporal analysis of specific conditions. Users can visualise trends for selected rejection types at farm level and explore how these change over time.

This section also includes automated alarm detection using statistical monitoring approaches, including Holt–Winters forecasting and CUSUM-based methods. These alerts highlight unusual patterns or potential emerging issues in carcass rejection data, supporting early detection and investigation.

For more information, please visit the FAQ section.

FAQ

General Queries

1. What constitutes Full Rejection?

Full Rejections refer to instances where the entire carcase of an animal is deemed unsuitable for human consumption.

2. What defines Partial Rejections?

Partial Rejections occur when specific parts of a pig’s carcase are deemed unfit for human consumption. Multiple partial rejections can be recorded for a single pig (e.g. hock and tail).

3. Can I upload any dataset into AbattoirInspect?

No. The dataset must follow the required structure and formatting described in the Upload Data tab. Incorrect formatting may result in upload errors or incomplete visualisations.

4. What should I do if my data upload fails?

Check that all required columns are present, correctly named, and formatted according to the template provided. If issues persist, test using the sample dataset to confirm system functionality before re-uploading.

5. Does AbattoirInspect store my uploaded data?

No. Uploaded data is only used within the current session for analysis and is not permanently stored by the system.

National Overview and My Farm(s) Tabs

6. How are the percentages in the bar plots calculated in the Basic tab?

Since the app displays counts and does not include slaughter population denominators, percentages are calculated as: (Number of rejections in a category per year / total rejections per year within the selected range) × 100.

7. How do the statistics in the My Farm(s) tab differ from the National Overview tab?

My Farm(s) statistics are based on user-selected farms and filters (e.g. date range, quarter). National Overview statistics are calculated across the full dataset within the selected time period.

8. Why do I see differences between National Overview and My Farm(s) results?

The National Overview includes all farms in the dataset, while My Farm(s) reflects only selected farms. Differences in scale and filtering will affect totals and percentages.

9. Are Top 5 farms and Top 10 producers calculated based on filters?

Yes. Rankings are calculated dynamically based on the selected date range and filters.

10. What is the purpose of the Benchmark plot?

The Benchmark plot compares selected farms against each other and against the national average for the top five conditions, enabling performance assessment and trend identification.

11. Why are there so many rejection types in my dataset?

The number of rejection types depends entirely on the dataset uploaded. If the list is extensive, users are encouraged to clean or group similar rejection categories to improve interpretability.

12. Can I compare my farm against a specific group of farms?

Yes. Multiple farms can be selected in the My Farm(s) tab to create custom comparison groups.

13. Why are some farms missing from the selection list?

Only farms present in the uploaded dataset and correctly formatted will appear in selection menus.

Time-Series and Statistics

14. Why are three years of data required for time-series analysis?

The system uses two years of data for model training and one year for evaluation, enabling more robust detection of trends and anomalies.

15. Can I change the training and evaluation period?

Yes. In Expert mode, users can adjust the number of years used for training and evaluation separately.

16. What is Holtwinters?

Holtwinters algorithm also known as Exponential Smoothing, is primarily used for time series forecasting. The algorithm has three main components: level (a baseline estimate), trend, and seasonality. These components are updated using weighted averages to account for changes in the data over time

17. What is CUSUM?

CUSUM is a control chart technique used for change detection or anomaly detection. It identifies shifts in the mean or variance of a process over time.CUSUM works by calculating a cumulative sum of deviations from a target value. When the cumulative sum exceeds a certain threshold, it signals a potential shift or change.

18. What is EWMA?

An EWMA (Exponentially Weighted Moving Average) chart is a statistical process control tool used to monitor small shifts in a process mean over time. It assigns exponentially decreasing weights to past data points, giving more importance to recent observations while still considering historical data

19. What does an alarm mean in the time-series plots?

An alarm indicates that observed values deviate significantly from expected behaviour based on statistical thresholds. It signals a potential anomaly requiring further investigation.

20. Should alarms be interpreted as confirmed disease events?

No. Alarms are early warning indicators only and should be interpreted alongside expert veterinary or epidemiological assessment.

Interpretation and Limitations

21. Are results adjusted for the number of animals slaughtered?

No. The application currently presents counts and derived percentages based on available rejection records and does not adjust for slaughter population size.

22. Can AbattoirInspect be used for regulatory reporting?

No. The tool is designed for surveillance, monitoring, and decision support. It should not be used as a standalone regulatory reporting system.

Support and Guidance

23. Where can I find more information about how to use the tool?

A short explainer video is available within the Help section of the application, providing a guided overview of the main features and tabs. For more information, feel free to contact SRUC staff – you can find contact details in the corresponding section.