Using Machine Learning to Predict Swine Movements within a Regional Program to Improve Control of Infectious Diseases in the US.

Between-farm animal movement is one of the most important factors influencing the spread of infectious diseases in food animals, including in the US swine industry. Understanding the structural network of contacts in a food animal industry is prerequisite to planning for efficient production strategies and for effective disease control measures. Unfortunately, data regarding between-farm animal movements in the US are not systematically collected and thus, such information is often unavailable. In this paper, we develop a procedure to replicate the structure of a network, making use of partial data available, and subsequently use the model developed to predict animal movements among sites in 34 Minnesota counties. First, we summarized two networks of swine producing facilities in Minnesota, then we used a machine learning technique referred to as random forest, an ensemble of independent classification trees, to estimate the probability of pig movements between farms and/or markets sites located in two counties in Minnesota. The model was calibrated and tested by comparing predicted data and observed data in those two counties for which data were available. Finally, the model was used to predict animal movements in sites located across 34 Minnesota counties. Variables that were important in predicting pig movements included between-site distance, ownership, and production type of the sending and receiving farms and/or markets. Using a weighted-kernel approach to describe spatial variation in the centrality measures of the predicted network, we showed that the south-central region of the study area exhibited high aggregation of predicted pig movements. Our results show an overlap with the distribution of outbreaks of porcine reproductive and respiratory syndrome, which is believed to be transmitted, at least in part, though animal movements. While the correspondence of movements and disease is not a causal test, it suggests that the predicted network may approximate actual movements. Accordingly, the predictions provided here might help to design and implement control strategies in the region. Additionally, the methodology here may be used to estimate contact networks for other livestock systems when only incomplete information regarding animal movements is available.

Potential role of noncommercial swine populations in the epidemiology and control of porcine reproductive and respiratory syndrome virus.

OBJECTIVE: To assess the role of noncommercial pigs in the epidemiology of porcine reproductive and respiratory syndrome (PRRS) virus. DESIGN: Seroepidemiologic study and survey study. ANIMALS: 661 pigs from which blood samples were collected at slaughter and 32 pigs from which blood samples were collected longitudinally. PROCEDURES: Spatial databases of commercial farms and 4-H participation were evaluated by use of commercial geographic information systems software. Information on disease knowledge and management methods of 4-H participants was obtained by mail survey and personal interview. Serum samples for antibody testing by PRRS ELISA were obtained from pigs at slaughter or at county fairs and on farms. RESULTS: Participation in 4-H swine programs was geographically associated with commercial swine production in Minnesota, and 39% of 4-H participants reared pigs at locations with commercial pigs. High seroprevalence at fairs (49%; range, 29% to 76%) and seroconversion after fairs indicated that PRRS virus exposure was common in pigs shown by 4-H participants and that transmission could occur at fairs. CONCLUSIONS AND CLINICAL RELEVANCE: The small swine population shown by 4-H members (estimated 12,000 pigs) relative to the population of commercial swine in Minnesota (estimated 6.5 million pigs) suggested the former overall was likely of minor importance to PRRS virus epidemiology at present. However, the relative frailty of knowledge of biosecurity practices, evidence that PRRS virus exposure was frequent, common intentions to show pigs at multiple events, and often close interactions with commercial herds suggested that the 4-H community should be involved in regional efforts to control PRRS.

Epidemiologic assessment of porcine circovirus type 2 coinfection with other pathogens in swine.

OBJECTIVE: To identify important pathogens and characterize their serologic and pathologic effects in porcine circovirus type 2 (PCV2)-infected pigs in relation to pig age and type of swine production system. DESIGN: Cross-sectional study. ANIMALS: 583 conventionally reared pigs. PROCEDURES: 3- (n = 157), 9- (149), 16- (152), and 24-week-old (125) pigs from 41 different 1-, 2-, and 3-site production systems (5 pigs/age group/farm) were euthanized and necropsied. Pigs with and without PCV2 infection were identified (via PCR assay); infection with and serologic responses to other pathogens and pathologic changes in various tissues (including lungs) were assessed. Logistic regression models were constructed for effects overall and within each age group and type of production system. RESULTS: Compared with PCV2-negative pigs, PCV2-positive pigs were more likely to have swine influenza virus (SIV) type A and Mycoplasma hyopneumoniae infections and sample-to-positive (S:P) ratios for SIV H1N1 from 0.50 to 0.99; also, PCV2-positive pigs had higher serum anti-porcine reproductive and respiratory syndrome virus (PRRSV) antibody titers and more severe lung tissue damage. Infection with SIV (but lower SIV H1N1 S:P ratio) was more likely in 3-week-old PCV2-positive pigs and evidence of systemic disease was greater in 16-week-old PCV2-positive pigs than in their PCV2-negative counterparts. By site type, associations of coinfections and disease effects between PCV2-positive and -negative pigs were greatest in 3-site production systems. CONCLUSIONS AND CLINICAL RELEVANCE: In PCV2-positive pigs, coinfections with SIV, M. hyopneumoniae, and PRRSV are important, having the greatest effect in the early to late nursery phase and in 3-site production systems.

Assessment of the adequacy of available spatial data of the demographics of swine populations in Minnesota.

Reliable data on the demographics of animal populations are essential for effective planning and execution of disease control programmes. To document the spatial distribution of different swine populations in Minnesota, the authors evaluated four sources of data from separate entities that maintain data independently and for different purposes. Although the total numbers of swine sites (aggregated at the county level) were significantly correlated among datasets, analysis of spatial clustering patterns demonstrated regional biases among the datasets. We used current, field verified data of farm locations in two counties to identify and quantify inaccuracies in two databases recording individual farm locations. The proportion of omitted or erroneous farm sites and the magnitude of positional inaccuracies were sufficient to limit the utility of available data for analytic purposes or for disease control efforts. There is a clear need for more current and accurate demographic data to underpin industry or government initiatives to control swine disease in Minnesota. Current efforts under the National Animal Identification System may address this concern.

Real-time disease surveillance tools for the swine industry in Minnesota.

The ultimate challenge for integrating geographic information systems (GIS) into swine veterinary activities in the United States is to develop systems that deliver effective decision support to practising veterinarians seeking to control hyper-endemic viral diseases such as porcine reproductive and respiratory syndrome (PRRS). Through collaboration with specialist swine veterinarians and state agencies we have developed two tools designed to enable real-time surveillance for swine diseases, based on capture of veterinary clinical data via the internet. The first is an ArcIMS-based Web-mapping application that enables authorised veterinarians to explore high resolution maps of swine premises throughout the state of Minnesota. This is coupled with a database that enables veterinarians to edit or update farm attribute data, including disease status via the internet. The system is generic so that novel or emerging disease syndromes can be incorporated. The second application is an extension to the Rapid Syndrome Validation Project-Animals which was developed for beef cattle to detect changes in disease incidence through recording of patterns of symptoms. The rationale behind these initiatives is that recent technological advances in relevant disciplines provide opportunities for innovation in surveillance that can enhance the capacity of the industry to combat its most pressing disease challenges in a more efficient manner. The acceptance of these novel tools by industry will assist veterinarians and producers in managing common disease problems, and hopefully serve to diminish the historic reluctance of producers to share information about disease status.