Available online at www.sciencedirect.com
ScienceDirect Procedia Food Science 7 (2016) 101 – 104
9th International Conference on Predictive Modelling in Food
FoodChain-Lab: Tracing software supporting foodborne disease outbreak investigations Armin A. Weiser *, Christian Thoens, Alexander Falenski, Bernd Appel, Matthias Filter and Annemarie Kaesbohrer Federal Institute for Risk Assessment, Berlin, Germany
Abstract In case of foodborne disease outbreaks, rapid identification of the causative food product is essential, since the medical and economic damages grow with the duration of the outbreak. Therefore and based on the experiences from recent foodborne disease outbreaks a free, open source software called FoodChain-Lab has been developed. It evolved from a supply chain data visualization and analysis tool into a comprehensive toolbox for data management, data enrichment, visualization, data analysis and interactive reasoning. FoodChain-Lab is applicable in food- or feed-borne disease outbreak investigations as well as in exposure assessment tasks related to feed or food supply chains.
© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license © 2015 The Authors. Published by Elsevier Ltd. (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of Department of Food Science, Faculty of Food Engineering, University of Campinas. Peer-review under responsibility of Department of Food Science, Faculty of Food Engineering, University of Campinas. Keywords: food/feed supply chain, foodborne infections, outbreaks, tracing analysis, open source software, GIS
1. Introduction In case of foodborne disease outbreaks, rapid identification of the causative food product(s) is essential. The approach usually applied in complex scenarios is to reconstruct the relevant food chains and distribution networks. This network reconstruction and its thorough analysis is a time-consuming and labor intensive effort.
* Corresponding author. Tel.: +49-30-18412-2118; fax: +49-30-18412-2952. E-mail address:
[email protected]
2211-601X © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of Department of Food Science, Faculty of Food Engineering, University of Campinas.
doi:10.1016/j.profoo.2016.02.097
102
Armin A. Weiser et al. / Procedia Food Science 7 (2016) 101 – 104
Beginning with the outbreak caused by Shiga toxin-producing Escherichia coli O104:H4 in Germany in 2011, the Federal Institute for Risk Assessment (BfR) started to develop a GPL-licensed open-source software tool called “FoodChain-Lab” which supports trace-back and trace-forward analysis of suspicious feed or food items along the supply chains. This ad-hoc development led to the identification of a specific lot of fenugreek seeds imported from Egypt as the most likely source of contamination1. Since then FoodChain-Lab has been further developed and successfully applied and tested in other outbreak investigations, such as an outbreak of norovirus gastroenteritis in September/October 2012 in Germany2 and an unusual increase in hepatitis A cases starting in 2013 within several European countries3. These real world outbreak situations revealed that there is a high demand of such software systems capable of supporting investigation on supply chains. The objective of this research was therefore to create a free, open-source and user-friendly software resource for public health experts. Definitions Station represents all kinds of food receivers and deliverers including final recipients (e.g. companies, persons) Trace stands for the supply chain paths a contamination can take (also respecting mixing and splitting events)
2. Results 2.1. Data Integration FoodChain-Lab provides scenario-adapted Excel templates allowing users to easily collect all necessary information and import it into its internal database. Within the database data can be validated, corrected and updated. Plausibility checks and similarity searches ensure high data-quality. 2.2. Data Analysis and Visualization A major application area of FoodChain-Lab is the analysis and visualization of food tracing information. This is accomplished by constructing and visualizing intuitive food delivery network graphs. These graphs can be analyzed interactively in FoodChain-Lab. If geographical information (latitude and longitude data) is not present FoodChain-Lab is able to generate that information from address information associated to stations. Geographical coordinates are necessary for GIS-based visualizations and analyses. A key functionality of the software is to search for a common station or delivery that has connections to all (or the majority of the) defined (outbreak) cases. FoodChain-Lab automatically computes a so-called “tracing score” for all stations and deliveries. The higher the score of a station/delivery, the more likely it is that a contamination of a commodity at this station/delivery can explain the selected cases. The tracing score is calculated using the following formula:
¦ ¦
n
Score(s i )
j 1 n
w j t ij
j 1
wj
where si is the i-th station or delivery, wi is the weight of the j-th station or delivery, tij has a value of 1, if there is a trace between si and sj and a value of 0 otherwise, n is the total number of stations and deliveries: The value of the tracing score Score(si) is between zero and one. FoodChain-Lab provides a visualization of that score that supports the fast identification the most important stations.
Armin A. Weiser et al. / Procedia Food Science 7 (2016) 101 – 104
Other useful features of FoodChain-Lab are x Observing: this feature shows the whole forward and backward trace of a user-defined set of stations/deliveries. x Cross-contamination: this feature allows for the simulation of cross-contamination events in products. x Regional analysis: this feature allows for automated simulation of regional effects in the investigated scenario, e.g. if there regions using contaminated irrigation water from the same origin. An impression of the main view provided by the software is given in Fig. 1
Fig. 1. Interactive trade network visualization: network graph (left) and GIS map view (right). Stations on the left and on the right are identical and always synchronized. In contrast to the GIS view, it is possible to manually reposition or automatically group stations that are connected via deliveries in the network view. Coloring is highly customizable. The size of the stations correlates usually with the tracing score.
2.3. Software Architecture FoodChain- Lab is free and licensed under the GNU General Public License. It has been implemented as a modular extension to the open source data analytics platform Konstanz Information Miner (KNIME)4. KNIME enables visual assembly of data analysis workflows. All functionalities of FoodChain-Lab have been written in JAVA programming language. Extensive descriptions explaining functionality and the user interface in detail has been directly added to the software. The installation guide, source code, example workflows, video tutorial, sample data and a ticket system are available via http://foodrisklabs.bfr.bund.de.
3. Conclusion FoodChain-Lab is a software tool designed from the beginning to support governmental outbreak investigations in food related outbreak situations. It has been challenged with real world data and data exchange formats resulting in the development of a sophisticated data management infrastructure capable of assuring high data quality and integrity while maintaining a high level of flexibility with respect to any data pre- and post-processing task. It allows
103
104
Armin A. Weiser et al. / Procedia Food Science 7 (2016) 101 – 104
for real-time visual analysis and simulation of hypotheses on the cause of food- or feed-borne disease outbreaks including geographical aspects. It is freely available and ready-to-use for everyone. Acknowledgements This work has been funded by the German national research project SiLeBAT, research grant 13N11202, and the Federal Institute for Risk Assessment, Germany. References 1. Weiser AA, Gross S, Schielke A, Wigger J-F, Ernert A, Adolphs J, et al. Trace-back and trace-forward tools developed ad hoc and used during the STEC O104:H4 outbreak 2011 in Germany and generic concepts for future outbreak situations. Foodborne Pathog Dis. 2013;10(3):263–9. 2. Anonymous. Großer Gastroenteritis-Ausbruch durch eine Charge mit Noroviren kontaminierter Tiefkühlerdbeeren in Kinderbetreuungseinrichtungen und Schulen in Ostdeutschland. Epidemiol Bull. 2012;41:414–7. 3. European Food Safety Authority. Tracing of food items in connection to the multinational hepatitis A virus outbreak in Europe. 2014;12(9):1–186. 4. Berthold MR, Cebron N, Dill F, Gabriel TR, Kötter T, Meinl T, et al. KNIME - the Konstanz information miner. ACM SIGKDD Explor Newsl. 2009; 16;11(1):26.
