Dr. Kimberly VanderWaal - Can We Predict Outbreaks of PRRS and PED Viruses?
•Download as PPTX, PDF•
2 likes•66 views
Can We Predict Outbreaks of PRRS and PED Viruses? - Dr. Kimberly VanderWaal, from the 2018 Allen D. Leman Swine Conference, September 15-18, 2018, St. Paul, Minnesota, USA. More presentations at http://www.swinecast.com/2018-leman-swine-conference-material
Recommended
Recommended
More Related Content
What's hot
What's hot (12)
Similar to Dr. Kimberly VanderWaal - Can We Predict Outbreaks of PRRS and PED Viruses?
Similar to Dr. Kimberly VanderWaal - Can We Predict Outbreaks of PRRS and PED Viruses? (20)
More from John Blue
More from John Blue (20)
Recently uploaded
Recently uploaded (20)
Dr. Kimberly VanderWaal - Can We Predict Outbreaks of PRRS and PED Viruses?
- 1. Can we predict outbreaks of PRRS and PED viruses? Kim VanderWaal, PhD Rahul Bhojwani, Igor Paploski, Carles Vilalta, Andres Perez, Gustavo Machado,Cesar Corzo September 17, 2018
- 2. Acknowledgements • Participating companies • MSHMP team – Juan Sanhueza – Carles Vilalta – Emily Geary – Paulo Fioravante Gustavo Machado now faculty at NCSU Igor Paploski Post-doc Rahul Bhojwani Data Science MSc
- 3. Morrison Swine Health Monitoring Project • Surveillance for endemic swine diseases in the US • Industry participation – ~45% of U.S. sow population • Long-term epidemiological monitoring & research • Data – Weekly infection status of sow farms for PRRS and PED – Locations of farms
- 4. Components of infection risk Spatially independent • Connectivity and structure of networks influence disease spread
- 5. Components of infection risk Spatially dependent Spatially independent
- 6. Data capture Data capture Wind Etc. Spatial data Weather Pig movements Visualize & Analyze Interpret & Report MSHMP data Precipitation / Humidity Quantify and dynamically predict the risk of outbreaks in sow farms Forecasting the risk of outbreaks
- 7. Neighborhood effects: • Outbreaks in neighboring farms • Vaccination • Animal movements • Status of origin • Micro-environment (climate, land use, vegetation, tree cover) PED outbreak? Neighborhood effects on risk
- 8. Neighborhood effects on risk Machado et al, In review Machine learning Data Model building Model testing Total farms = 3269 Sow farms = 319 PRRS breaks = 321 PED breaks = 458
- 10. Variable importance: PEDv • More pigs moved into neighborhood Higher risk • Infection status of source • More vegetation Generally low risk
- 11. Week 4 Week 14 PPV=76% NPV=99% Sens=18%
- 13. PRRS genetic groups Igor Paploski, post-doc Lineage present Lineage absent No data Lineage 8* *Network k-test for clustering of cases, p < 0.001
- 15. Limitations • Not all PRRS viruses are the same • Data availability and timeliness – Movement data – Infection data • Missing pieces – Biosecurity and vaccination – Alternative modes of transmission?
- 16. Conclusions • A neighborhood-based approach captures disease risks associated with long-distance animal movement and local spatial dynamics • Dynamics may depend on type of PRRS • Foundation for dynamic forecasting of risk
- 17. Thank you!
Editor's Notes
- 2018_edin talk was 50 minutes and 54 slides, including title etc. This needs to cut to 27 slides
- This project was done using databases upkept by the MSHMP for surveillance of swine diseases in the US. Databases consist of.
- Dates & Number head (~2 years) - 100,000 movements, accounting for 20 million pigs - 25,000 pigs daily - 125 truckloads daily
- Combination of long distance and short distance processes, with environemtnal factos effecting local spread
- So let’s move back to pigs. The overarching goal of our work is to quantify and dynamically predict the risk….
- Chapter 3. I had the idea to apply machine learning neighborhood affects, which is an example of the type or creativity that I bring to research. Lucky enough to work with the team to implement it 125 trucks / 25,000 pigs daily
- Figure. Molecular Phylogenetic analysis by Maximum Likelihood method The evolutionary history was inferred by using the Maximum Likelihood method based on the Tamura-Nei model [1]. The tree with the highest log likelihood (-15756.04) is shown. Initial tree(s) for the heuristic search were obtained automatically by applying Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances estimated using the Maximum Composite Likelihood (MCL) approach, and then selecting the topology with superior log likelihood value. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. The analysis involved 1878 nucleotide sequences. All positions containing gaps and missing data were eliminated. There were a total of 449 positions in the final dataset. Evolutionary analyses were conducted in MEGA7 [2].
- Often overlooked in entowrk anlayis , which advance disease surveillance and control for endemic swine pathogens in the United States.