Thesis thoughts

What is Johne’s Disease?

  • Johne’s disease is a disease of ruminants (think cows, goats, sheep, deer).
  • It is caused by the bacteria Mycobacterium avium subspecies paratuberculosis (MAP).
  • Infection in dairy herds happens early in life, usually animals less than 6 months of age.
  • It typically takes years before infected animals show signs of illness.
  • It is slowly progressive and untreatable (unresponsive to antibiotics and deworming).
  • Clinical signs are weight loss despite a good appetite, increased thirst, intermittent to persistent diarrhea.
  • The disease is considered an “iceberg disease” meaning that when we see cows sick with Johne’s disease they represent only the very tip of the true number of infections. Many more are subclinically infected (infected but not showing illness yet).

What is the issue with Johne’s disease in Ontario, in Canada, and worldwide?

  • Prevalence has been increasing with some researchers estimating 90% of American dairy farms are infected while approximately 46% of Canadian herds are believed to be infected.
  • There have been studies who have found associations between the presence of MAP (the bacteria that causes Johne’s) and immune disease in humans such as Crohn’s, Hashimoto’s disease and Type 1 diabetes.
    • Association DOES NOT mean causation, these studies have suggested there is some sort of genetics at play in these associations as well.
  • Animals in the clinical stages of Johne’s have reduced milk production, have a shorter production lifespan and have reduced value at slaughter due to decreased body condition.
  • There is suspicion that these sick animals also suffer from reduced reproductive ability and are susceptible to other diseases such as mastitis.
  • Apart from animal welfare, economic losses range from $28 million USD annually in Canada to $198 million USD annually in the American dairy industry.

How do we typically try to control Johne’s?

  • As there are no known treatments, control of Johne’s is often based on testing to detect infected animals and removing them from the herd
  • This testing and removal/culling is often paired with changing/altering management practices to eliminate risk of either introducing it into the herd (if they don’t have it yet) or reducing the risk of it being transmitted to other cows.

What challenges do we face?

  • Biggest challenge is that Johne’s infected farms have many subclinical animals.
  • Subclinical animals are difficult to detect and diagnose.
  • These subclinical animals are able to spread more of the bacterium and infect more animals making more subclinical animals.
  • The time between infection (<6 months of age) and when they start to show clinical signs of illness (>2 years of age) creates a long lag period

What are producers saying?

  • Very little has been said about the disease in media/ government/ through industry publications, is it still an issue?
  • Know they have it in their herd but difficulties identifying and stopping its spread is frustrating, have resigned to managing the best they can
  • Need support from the industry to help with testing and knowing when and how to test their herd

What can machine learning offer to Johne’s research?

  • Random Forest algorithms offer moderate performance when predicting follow up test results
  • When selecting the top 25% of cows predicted by the model we can capture ~75% of the positive animals
  • Machine learning algorithms are difficult to interpret the ‘inner workings’, classification/decision tree may be an alternative procedure
  • Collection of detailed data would allow for further exploration of newer data analysis techniques