Experiences of a Risk Assessment Authority with KNIME - based Applications

Transcription

1 FEDERAL INSTITUTE FOR RISK ASSESSMENT Experiences of a Risk Assessment Authority with KNIME - based Applications Matthias Filter Federal Institute for Risk Assessment, Germany

2 Outline - What does a risk assessment authority do? - KNIME-based applications at BfR - Plans and Challenges Matthias Filter, 7th KNIME UGM, , Zürich Page 2

3 Federal Institute for Risk Assessment Established 1 November 2002 (its predecessor founded in 1876) Annual budget: appr. 65 Mio Research budget: appr. 6 Mio BfR staff: Locations: appr. 750 employees (550 working in the scientific field) Berlin (Marienfelde, Jungfernheide) Main areas of work: Feed and Food Safety Safety of Substances and Preparations Safety of Consumer Products Risk-Communication etc. Matthias Filter, 7th KNIME UGM, , Zürich Page 3

4 Main areas of work Microbiological and substance-chemical safety and assessment of foods, Safety and assessment of substances (chemicals, plant protection products, biocides) and selected products (commodities, cosmetics, tobacco products, textiles and food packaging), Risk communication, Research Furthermore Elaboration of alternatives to animal experiments in which the safety of substances can be determined Matthias Filter, 7th KNIME UGM, , Zürich Page 4

5 Food Safety Institutions in the European Union Risk management Risk assessment Analytics and diagnostics research Matthias Filter, 7th KNIME UGM, , Zürich Page 5

6 Food Safety Institutions Germany EU Commission Matthias Filter, 7th KNIME UGM, , Zürich Page 6

7 Department Biological Safety Products in working focus: Food Pathogens in working focus: Salmonella, E.coli (faecal germs), Mycobacteria, Brucella, Campylobacter, Listeria, Yersinia, Staphylococci various parasites, like Trichinella or Echinococcus and viruses 10 National Reference Laboratories for various zoonoses pathogens Matthias Filter, 7th KNIME UGM, , Zürich Page 7

8 Example: E. coli O104:H4 outbreak, 2011 Largest outbreak by EHEC infection in Germany so far Cases: Setting: Causative food: Epidemiol. evidence: 3793 (2353 hospitalized, 53 death) Disseminated cases (restaurants, hotels, canteens, household) Sprouted fenugreek seeds Cohort study, trace-back and network analyses (delivery chains of sprouts and seeds) Matthias Filter, 7th KNIME UGM, , Zürich Page 8

9 Example: Norovirus outbreak, 2012 Largest food-borne outbreak in Germany so far Cases: (38 hospitalized) Setting: Disseminated cases (at least 390 affected facilities, almost exclusively schools and kindergartens) Causative food: Epidemiol. evidence: Microbiol. evidence: Imported frozen strawberries Case-control studies, trace-back investigations Detection of outbreak strains in the suspected lot of frozen strawberries Matthias Filter, 7th KNIME UGM, , Zürich Page 9

10 Food-borne outbreaks in the EU Year Outbreaks without identified foodstuff Outbreaks with identified foodstuff Source: vegetables and fruits Source: EFSA Matthias Filter, 7th KNIME UGM, , Zürich Page 10

11 BfR-RESEARCH: How do we cope with food-borne outbreaks in the context of global food chains? Matthias Filter, 7th KNIME UGM, , Zürich Page 11

12 Global food supply chains Increased complexity of risk / exposure assessments A. Matopoulos, M. Vlachopoulou, V. Manthou, B. Manos, (2007), Supply Chain Management. Vol. 12 Iss: 3, pp Supply chain system Spatial Baseline levels Contaminants Temporal FOOD Distribution FOOD Distribution FOOD Processing FOOD Processing FOOD Consumption FOOD Consumption Matthias Filter, 7th KNIME UGM, , Zürich Page 12

13 Our Vision A community-driven food safety knowledge base (containing data, models, tools) animal disease models Food processing and distribution human disease models Environmental factors Processing parameters Matthias Filter, 7th KNIME UGM, , Zürich Page 13

14 BfR- Solutions Predictive Microbiology Data Analysis and Visualization DB Spatio- Temporal Modeling Food Process Simulation Rind fleisch Nierenfett Vorschneiden Verpackungsmaterial Froster Kuttern Kuttern Füller Därme Lager Verschweißen Reiferaum Raucherzeu- gung Matthias Filter, 7th KNIME UGM, , Zürich Page 14

15 Real world example: Norovirus outbreak 2012 A total of cases, averagely 10 years old At least 390 affected facilities (cluster), almost exclusively schools and kindergartens Matthias Filter, 7th KNIME UGM, , Zürich Page 15

16 FoodChain-Lab: BfR KNIME node library supporting outbreak investigations Matthias Filter, 7th KNIME UGM, , Zürich Page 16

17 FoodChain-Lab: BfR KNIME node library supporting outbreak investigations Application during Norovirus outbreak Kitchen + Customers Info for each node/arrow available Importer + Customers Matthias Filter, 7th KNIME UGM, , Zürich Page 17

18 FoodChain-Lab: BfR KNIME node library supporting outbreak investigations Identification of Outbreak-Clusters Supply chain analysis: - Tracing backward : Clusters common Producer / Source - Tracing forward: Producer / Source further Clusters - Network analysis Risk oriented sampling in factories, further tracing backward/forward Tracing backward Source / Producer Distributor Cluster Tracing forward Weiser et al., 2013: 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 Matthias Filter, 7th KNIME UGM, , Zürich Page 18

19 BfROpenLab: KNIME node library supporting GIS visualization of data Get Coordinates with OpenStreetMap: Matthias Filter, 7th KNIME UGM, , Zürich Page 19

20 PMM-Lab / FoodProcess-Lab: KNIME node libraries to support quantitative microbial risk assessments (QMRA) Database / Tools for food / feed processing parameters FoodProcess-Lab Database / Tools for models on microbial growth / inactivation PMM-Lab Matthias Filter, 7th KNIME UGM, , Zürich Page 20

21 PMM-Lab Matthias Filter, 7th KNIME UGM, , Zürich Page 21

22 PMM-Lab DB integrated data management Matthias Filter, 7th KNIME UGM, , Zürich Page 22

23 PMM-Lab use case: model generation Matthias Filter, 7th KNIME UGM, , Zürich Page 23

24 PMM-Lab use case: predictions Matthias Filter, 7th KNIME UGM, , Zürich Page 24

25 Advantages of PMM-Lab (and other BfR KNIME nodes) All relevant information on model generation process saved (raw data, data modifications, model generation workflow, estimated models) Modular, easy to use software, FREE Integration with other popular modelling systems available: R, SAS, Excel Allows exchange of models and model generation workflows -> opportunity to provide best practice -> create transparency into modelling research Easy to deploy as Web Service (unfortunately NOT FREE) Matthias Filter, 7th KNIME UGM, , Zürich Page 25

26 Challenge: in order to create real impact => Solutions should be provided as Web Services Internet Open-Source software Google KNIME Server PCsoftw are Bing Service 1 Service N Intranet BfR Matthias Filter, 7th KNIME UGM, , Zürich Page 26

27 Example: Web Service for risk assessors Matthias Filter, 7th KNIME UGM, , Zürich Page 27

28 Example: Web Service for companies / risk managers Matthias Filter, 7th KNIME UGM, , Zürich Page 28

29 Vision: Web based community food safety model repository Matthias Filter, 7th KNIME UGM, , Zürich Page 29

30 1st step: Apply Systems Biology Markup Language (SBML) / Numeric Markup Language (NuML) as data exchange formats for Food Safety models and data Courtesy: Frank T. Bergmann (University of Heidelberg) Matthias Filter, 7th KNIME UGM, , Zürich Page 30

31 Plans: Consolidation of our developments / Development of product level like solutions (usability etc.) Experts Companies Other BfR Research product level!! Prototypes Basic research Communities Scientific publications, opensource software, networks New R&D projects Governments / Authorities (including BfR staff) Matthias Filter, 7th KNIME UGM, , Zürich Page 31

32 FEDERAL INSTITUTE FOR RISK ASSESSMENT Thankto theteam Christian Thöns, Armin A. Weiser, Alexander Falenski, Carolina Plaza Rodriguez, Bernd Appel, Annemarie Käsbohrer Thank you for your attention Matthias Filter Federal Institute for Risk Assessment Max-Dohrn-Str Berlin, GERMANY Tel Fax