Comparison of the dynamic networks of four equine boarding and training facilities.

This study aims to understand and compare the patterns of contact between horses in different equestine facilities in Ontario using radio-frequency identification tags. The ultimate goal is to assess the potential for disease transmission within and between these facilities.

Research Methodology

• The study used contact pattern data from 4 equine facilities, which were collected using radio-frequency identification tags (RFID).
• These data were used to form 7 static contact networks for each facility over 7 days.
• The researchers then tested the assumption of homogenous mixing. This is the assumption that every individual equine has an equal probability of coming into contact with any other. This assumption is often used in disease transmission models.
• The researchers used the Jaccard Similarity Index to identify repeating contacts throughout the study period.
• Network models based on survey data were used to assess whether they could predict recorded RFID contacts.
• The study also examined the centrality of nodes, representing individual horses, in the network. This involved looking at the node degree (the number of contacts a horse has), and node strength (the importance of a horse in the network), along with the consistency of these measurements over time.
• A correlation analysis was conducted to determine the relationship between node degree and strength.
• The researchers used an exponential random graph model to measure the relationship between the distance of horse stalls and network edge formation.

Research Findings

• The study found that neither the daily nor the combined weekly networks showed homogenous mixing at the facility level.
• The Jaccard Similarity Index showed that 11-62% of contacts were repeated throughout the study.
• The survey-based network model was able to predict 8.7-79.6% of the RFID-recorded contacts.
• The normalized node degree ranged from 0.0 to 0.96, with an average of 0.31. The node strength ranged from 0 to 1, with an average of 0.38.
• The study found a significant but weak positive correlation between node degree and strength.
• The betweenness centrality, a measure of a node’s centrality in the network, revealed a range from 0.00 to 1.00 and a mean of 0.11.
• The distance between horse stalls did not significantly influence network structure across the facilities.

Conclusion

• The study concluded that equine contact patterns are non-homogenous, and there’s a wide range of centrality measures. This indicates that empirical data is essential to understand processes such as disease spread potential within equine populations.
• The study proves that in situations where there’s little to no data, contact networks can be inferred from observational data.
• The findings serve as a starting point for characterizing equine contact networks in Ontario.