Interpretable machine learning applied to on-farm biosecurity and porcine reproductive and respiratory syndrome virus

A scoping review of knowledge, attitudes, and practices in swine farm biosecurity in North America

Pork is one of the most popular consumer meat choices globally, second to poultry. In the past two decades, the rising demand in pork, has seen pig farming move toward intensive farming methods, characterized by high pig densities which is a risk for swift spread of disease necessitating proper and strict biosecurity adherence to facilitate disease-free conditions and business continuity. North America is the second largest pig producer globally. We conducted a review of available peer-reviewed original publications to scope for available data on the knowledge, attitudes, perceptions, and practices concerning biosecurity among swine producers in North America from the year 2011 to 2022 using the PRISMA-SCr guidelines. Out of the 323 papers that fit our search criteria, we present insights from the 18 papers that were relevant to our study. We summarize key findings on biosecurity practices and propose critical practices for biosecurity adherence. We also present our findings on the complexities that influence producers' adoption of biosecurity plans and note variations in biosecurity strictness between states and how these are influenced by farm size and perceived disease risk. In conclusion, this review highlights the need for updated assessments of biosecurity practices, leveraging technology particularly machine learning, for risk assessment, and acknowledges the role that demographics and risk perception play in biosecurity adoption. Ultimately, effective biosecurity measures are imperative for safeguarding North American swine production systems against disease threats especially foreign animal diseases like the African swine fever (ASF), foot and mouth disease (FMD) and classical swine fever.

Understanding Post-Translational Modifications in Porcine Reproductive and Respiratory Syndrome Virus Infection

Porcine reproductive and respiratory syndrome virus (PRRSV) is a highly contagious virus affecting pigs with significant impacts to the swine industry worldwide. This review provides a comprehensive understanding of post-translational modifications (PTMs) associated with PRRSV infection. We discuss the various types of PTMs, including phosphorylation, ubiquitination, SUMoylation, acetylation, glycosylation, palmitoylation, and lactylation, that occur during PRRSV infection. We emphasize how these modifications affect the function and activity of viral proteins, thereby influencing virus replication, assembly, and egress. Additionally, we delve into the host cellular responses triggered by PRRSV, particularly the PTMs that regulate host signaling pathways and immune responses. Furthermore, we summarize the current understandings of how PTMs facilitate the ability of virus to evade the host immune system, enabling it to establish persistent infections. Finally, we address the implications of these modifications in the development of novel antiviral strategies and the potential for exploiting PTMs as therapeutic targets. This review highlights the significance of PTMs in shaping viral pathogenicity and host antiviral mechanisms and provides valuable insights for future research aimed at developing effective interventions against PRRSV infections.

Molecular mechanism of autophagy in porcine reproductive and respiratory syndrome virus infection

Porcine reproductive and respiratory syndrome virus (PRRSV), a significant pathogen affecting the swine industry globally, has been shown to manipulate host cell processes, including autophagy, to facilitate its replication and survival within the host. Autophagy, an intracellular degradation process crucial for maintaining cellular homeostasis, can be hijacked by viruses for their own benefit. During PRRSV infection, autophagy plays a complex role, both as a defense mechanism of the host and as a tool exploited by the virus. This review explores the current understanding of the molecular mechanisms underlying autophagy induction under PRRSV infection, its impact on virus replication, and the potential implications for viral pathogenesis and antiviral strategies. By synthesizing the latest research findings, this article aims to enhance our understanding of the intricate relationship between autophagy and PRRSV, paving the way for novel therapeutic approaches against this swine pathogen.

Comparison of Machine Learning Tree-Based Algorithms to Predict Future Paratuberculosis ELISA Results Using Repeat Milk Tests

Machine learning algorithms have been applied to various animal husbandry and veterinary-related problems; however, its use in Johne's disease diagnosis and control is still in its infancy. The following proof-of-concept study explores the application of tree-based (decision trees and random forest) algorithms to analyze repeat milk testing data from 1197 Canadian dairy cows and the algorithms' ability to predict future Johne's test results. The random forest models using milk component testing results alongside past Johne's results demonstrated a good predictive performance for a future Johne's ELISA result with a dichotomous outcome (positive vs. negative). The final random forest model yielded a kappa of 0.626, a roc AUC of 0.915, a sensitivity of 72%, and a specificity of 98%. The positive predictive and negative predictive values were 0.81 and 0.97, respectively. The decision tree models provided an interpretable alternative to the random forest algorithms with a slight decrease in model sensitivity. The results of this research suggest a promising avenue for future targeted Johne's testing schemes. Further research is needed to validate these techniques in real-world settings and explore their incorporation in prevention and control programs.

A discrete-time survival model for porcine epidemic diarrhoea virus

Since the arrival of porcine epidemic diarrhea virus (PEDV) in the United States in 2013, elimination and control programmes have had partial success. The dynamics of its spread are hard to quantify, though previous work has shown that local transmission and the transfer of pigs within production systems are most associated with the spread of PEDV. Our work relies on the history of PEDV infections in a region of the southeastern United States. This infection data is complemented by farm-level features and extensive industry data on the movement of both pigs and vehicles. We implement a discrete-time survival model and evaluate different approaches to modelling the local-transmission and network effects. We find strong evidence in that the local-transmission and pig-movement effects are associated with the spread of PEDV, even while controlling for seasonality, farm-level features and the possible spread of disease by vehicles. Our fully Bayesian model permits full uncertainty quantification of these effects. Our farm-level out-of-sample predictions have a receiver-operating characteristic area under the curve (AUC) of 0.779 and a precision-recall AUC of 0.097. The quantification of these effects in a comprehensive model allows stakeholders to make more informed decisions about disease prevention efforts.

Modelling and assessing additional transmission routes for porcine reproductive and respiratory syndrome virus: Vehicle movements and feed ingredients

Accounting for multiple modes of livestock disease dissemination in epidemiological models remains a challenge. We developed and calibrated a mathematical model for transmission of porcine reproductive and respiratory syndrome virus (PRRSV), tailored to fit nine modes of between-farm transmission pathways including: farm-to-farm proximity (local transmission), contact network of batches of pigs transferred between farms (pig movements), re-break probabilities for farms with previous PRRSV outbreaks, with the addition of four different contact networks of transportation vehicles (vehicles to transport pigs to farms, pigs to markets, feed and crew) and the amount of animal by-products within feed ingredients (e.g., animal fat or meat and bone meal). The model was calibrated on weekly PRRSV outbreaks data. We assessed the role of each transmission pathway considering the dynamics of specific types of production (i.e., sow, nursery). Although our results estimated that the networks formed by transportation vehicles were more densely connected than the network of pigs transported between farms, pig movements and farm proximity were the main PRRSV transmission routes regardless of farm types. Among the four vehicle networks, vehicles transporting pigs to farms explained a large proportion of infections, sow = 20.9%; nursery = 15%; and finisher = 20.6%. The animal by-products showed a limited association with PRRSV outbreaks through descriptive analysis, and our model results showed that the contribution of animal fat contributed only 2.5% and meat and bone meal only .03% of the infected sow farms. Our work demonstrated the contribution of multiple routes of PRRSV dissemination, which has not been deeply explored before. It also provides strong evidence to support the need for cautious, measured PRRSV control strategies for transportation vehicles and further research for feed by-products modelling. Finally, this study provides valuable information and opportunities for the swine industry to focus effort on the most relevant modes of PRRSV between-farm transmission.

A comparison of explainable artificial intelligence methods in the phase classification of multi-principal element alloys

We demonstrate the capabilities of two model-agnostic local post-hoc model interpretability methods, namely breakDown (BD) and shapley (SHAP), to explain the predictions of a black-box classification learning model that establishes a quantitative relationship between chemical composition and multi-principal element alloys (MPEA) phase formation. We trained an ensemble of support vector machines using a dataset with 1,821 instances, 12 features with low pair-wise correlation, and seven phase labels. Feature contributions to the model prediction are computed by BD and SHAP for each composition. The resulting BD and SHAP transformed data are then used as inputs to identify similar composition groups using k-means clustering. Explanation-of-clusters by features reveal that the results from SHAP agree more closely with the literature. Visualization of compositions within a cluster using Ceteris-Paribus (CP) profile plots show the functional dependencies between the feature values and predicted response. Despite the differences between BD and SHAP in variable attribution, only minor changes were observed in the CP profile plots. Explanation-of-clusters by examples show that the clusters that share a common phase label contain similar compositions, which clarifies the similar-looking CP profile trends. Two plausible reasons are identified to describe this observation: (1) In the limits of a dataset with independent and non-interacting features, BD and SHAP show promise in recognizing MPEA composition clusters with similar phase labels. (2) There is more than one explanation for the MPEA phase formation rules with respect to the set of features considered in this work.