- New Flubird Database New Flubird Database – e e ub d ub d a abase a abase Platform for Data Exchange and Knowledge Platform for Data Exchange and Knowledge B ildi B ildi Building in Avian Influenza Surveillance Building in Avian Influenza Surveillance i A i i A i I fl I fl S S ill ill Staubach C 1 Mathey A 1 Kowalczyk S 1 Tubbs N 2 Wilking H 1 Richter S 1 Kranz P 1 Staubach C , Mathey A , Kowalczyk S , Tubbs N , Wilking H , Richter S , Kranz P , Hagemeijer W 2 , Harder T 1 , Conraths FJ 1 and the NFB consortium 3 1 Friedrich-Loeffler-Institut, Federal Research Institute of Animal Health, Germany 2 Wetlands International Headquarters, The Netherlands 3 Coordinated by Osterhaus A, Erasmus University Medical Center, The Netherlands
Outline : Background Background • New FluBird (Network for Early Warning of Influenza Viruses in (Network for Early Warning of Influenza Viruses in Migratory Birds in Europe) • Challenges • Challenges Database • Design / Technical aspects D i / T h i l t • „Walk through“ / Data flow • Integration of International Waterbird Census I t ti f I t ti l W t bi d C Outlook • Upcoming developments
New FluBird - Objectives: • Interdisciplinary approach • Integration of different data sources Integration of different data sources • Evidence based surveillance Better understanding of avian influenza ecology Basis for predictive modelling Basis for predictive modelling More effective risk assessment
Project participants:
Avian influenza (AI) in wild birds Reservoir of AIV: Birds of the genus Anseriformes and Anseriformes and Charadriiformes (water associated habitat) ) Isolation of all 16 HA and 9 NA HA- and 9 NA subtypes Virus are usually Virus are usually low pathogenic Genetic reassortment Anseriformes Charadriiformes [image removed] [image removed]
Transmission Cycle of AI Biotic and Animal contacts, faeces / water abiotic vectors LPAIV Free ranging Humans, Humans, vectors, animal contacts Intensive farms Mutation of H5/H7 f H5/H7 High economic Losses in the poultry sector Reservoir: wild water birds: HPAI – Avian Influenza LPAI H1 - H16 [image removed] [image removed]
[image removed]
Year-round prevalence of LPAI 7,00 600 6,00 5,00 4,00 3,00 2,00 1,00 , 0,00 1 2 3 4 5 6 7 8 9 10 11 12 Month
Challenges: Compatibility to external initiatives for possibility of data exchange (e.g. EC, Wetlands International) Ad Adapted database structure and upload interfaces t d d t b t t d l d i t f Code bridges, e.g. bird species codes (i.e. WBDB, EURING) Diverse and large user community Diverse and large user community User manager: flexible, decentralized user account administration User-based, configurable data access rights , g g Possibility of including “User groups”, e.g. External Advisory Board, EC, EFSA? I t Interactivity, Transparency, Integration ti it T I t ti User friendly software modules for data interaction Visualization via map server Visualization via map server
Database structure: • Table 1a/b: Laboratory results compatible with EC/CRL compatible with EC/CRL • Table 2: Bird observation questionnaire compatible with GAINS (e.g. census data) • Table 3: Table 3: Bird watching site description Bird watching site description compatible with GAINS (e.g. CSN tool) • Table 4: Bird observation missions T bl 4 Bi d b ti i i background data (addition to table 2)
Partner Partner Species EC, CRL Laboratories Scientific Institutes Ornithology Partner
User manager:
German surveillance database Number of Records per year and status of birds Number of Records per year and status of birds Active sampling g Passive sampling g Bird Pos. Pos. status alive status alive hunted hunted H5N1 dead H5N1 dead sick sick H5N1 H5N1 5.800 1.136 23.104 40 2006 0 316 2007 15.819 1.782 1 7.892 101 331 2008 14.454 2.883 0 4.744 35 0 2009 6.430 648 1 2.389 53 0
Report of all investigations of wild birds regarding AI in charge of the Federal States to the EC by FLI in charge of the Federal States to the EC by FLI
Data upload dialog: lab data
„Upload“ Files: XML-File ASCII text file
Data access and digestion: Mi i Mining of data on basis of complex queries f d t b i f l i User friendly interface to handle queries; stored for re-use Any database field can be selected as search criteria Any database field can be selected as search criteria - single filter conditions can be flexibly combined by ‘AND’ or ‘OR’ logical links g Linking of different data types based on shared criteria (e.g. spatial, temporal, species related) Output of data / Visual integration Predefined reports, automatically generated f Presentation of user defined queries in table view Visualization of query results in map server e g combined Visualization of query-results in map-server, e.g. combined with selected flyway map layer, census data, etc.
IWC sites in WP & SW Asia 1990 - 2007:
IWC data selection by area
IWC – Tabular presentation of count aggregates
Cygnus cygnus Median > 100
Integration of CLC data Corine Land Cover Corine Land Cover - 44 classes of land coverage - 1:100,000 mapping scale, minimum mapping unit 25 hectares Background data for Map Server - Aim: Visual integration by overlaying with e.g. lab result layers, g y y g g y ornithological data layers, etc. Coverage profiles per geographical / administrative unit - Appropriately grouped categories => coverage profiles - Aim: enable filtering of sample events based on adjoin environmental parameters environmental parameters - Example query: • Select all M-PCR positive samples from areas with X % surface covered by wetlands agriculture etc covered by wetlands, agriculture, etc.
NFB-DB CORINE Landcover data Filter Map Data records on NUTS level
Surveillance schemes: The target of the surveillance must be fixed! The target of the surveillance must be fixed! HPAI and LPAI surveillance must be different! Surveillance can then be optimized to achieve these p goals and reduce resources HPAI surveillance must focus on dead birds (public awareness, sampling in the breeding areas, cooperation li i th b di ti with ornithologists, mortality reporting etc. required) LPAI surveillance could focus on some geographical LPAI surveillance could focus on some geographical (“representative”) hotspots, species (e.g. high prevalence and H/N diversity), time periods, specimen, y) mallard sentinel stations (e.g. Globig et al., 2009, EID) Both sample schemes could also include a small proportions of the other without loosing optimization ti f th th ith t l i ti i ti potential
Outlook: Database Database Expansion of the currently 128,000 records of investigated wild birds in time and space p Online calculation of raw prevalence maps and maps where the estimate is corrected for the sampling error Automated threshold warnings by the database (changes in Automated threshold warnings by the database (changes in prevalences, appearance of new subtypes, etc.) Integration of dynamic data on wild bird movements g y
Seasonal movements to and from Germany & Denmark for Mallard ( Anas platyrhynchos ) Season mallards ringed and recorded in Denmark & Germany in December until February
Outlook: Database Database Expansion of the currently 128,000 records of investigated wild birds in time and space p Online calculation of raw prevalence maps and maps where the estimate is corrected for the sampling error Automated threshold warnings by the database (changes in Automated threshold warnings by the database (changes in prevalences, appearance of new subtypes, etc.) Integration of dynamic data on wild bird movements g y Analysis/Modelling Combined space-time analysis/modelling of all data sources to C bi d ti l i / d lli f ll d t t understand better the ecology of AI in wild birds Optimize surveillance in space and time incl. identification of p p hotspots
Distribution of the sample size per time and space Months Number of samples High : 245 Low : 0 Municipalities Municipalities
Analysis/Modelling y g Each bird i has the unknown probability π i that it is positive or 1. negative depending on area j the bird lives, on the time t and g p g j , species x , the covariates, α 1…n The parameter π i is modeled with a logistic model 2. π i log = µ + θ + ϕ + s + α + ... + α + x 1 j t t n i 1 − π i μ = intercept θ j θ j = spatial effect in the area j spatial effect in the area j = time effect on the time t ϕ t s t = seasonal effect for the whole region α α 1..n = variables e g regarding CLC high risk IWC = variables e.g. regarding CLC, high risk, IWC x = species ~
Recommend
More recommend
