Using animal‐mounted sensor technology and machine learning to predict time-to-calving in beef and dairy cows

G A Miller, MA Mitchell, ZE Barker, Katharina Giebel, Edward Codling, Jonathan Amory, C Michie, Chris Davison, Christos Tachtatzis, Ivan Andonovic, C-A Duthie

Research output: Contribution to journalArticle

Abstract

Worldwide, there is a trend towards increased herd sizes and the animal to stockman ratio is increasing within the beef and dairy sectors, thus the time available to monitoring individual animals is reducing. The behaviour of cows is known to change in the hours prior to parturition, e.g. less time ruminating and eating, and increased activity level and tail raise events. These behaviours can be monitored non-invasively using animal mounted sensors. Thus behavioural traits are ideal variables for the prediction of calving. This study explored the potential of two sensor technologies for their capabilities in predicting when calf expulsion should be expected. Two trials were conducted at separate locations: i) beef cows (n = 144) and (ii) dairy cows (n = 110). Two sensors were deployed on each cow: 1) Afimilk Silent Herdsman (SHM) collars monitoring time spent ruminating (RUM), eating (EAT) and the relative activity level (ACT) of the cow and 2) tail mounted Axivity accelerometers to detect tail-raise events (TAIL). The exact time the calf was expelled from the cow was determined by viewing closed-circuit television camera footage. Machine learning random forest (RF) algorithms were developed to predict the when calf expulsion should be expected using single sensor variables and by integrating multiple sensor data-streams. The performance of the models were tested by the Matthew’s Correlation Coefficient (MCC), the area under the curve and the sensitivity and specificity of predictions. The TAIL model was slightly better at predicting calving within a five hour window for beef cows (MCC = 0.31) than for dairy cows (MCC = 0.29). The TAIL+RUM+EAT models were equally as good at predicting calving within a five hour window for beef and dairy cows (MCC = 0.32 for both models). Combining data-streams from SHM and tail sensors did not substantially improve model performance over tail sensors alone therefor hour-by-hour algorithms for the prediction of the time of calf expulsion were developed using tail sensor data. Optimal classification occurred at two hours prior to calving for both beef (MCC = 0.29) and dairy cows (MCC = 0.25). This study has shown that tail sensors alone are adequate for the prediction of parturition and that the optimal time for prediction is two hours before expulsion of the calf.
Original languageEnglish
JournalAnimal
Early online date13 Jan 2020
DOIs
Publication statusFirst published - 13 Jan 2020

Fingerprint

artificial intelligence
beef cows
sensors (equipment)
calving
dairy cows
tail
calves
prediction
cows
ingestion
beef
televisions (equipment)
parturition
animals
monitoring
herd size
collars
cameras
dairies
taxonomy

Keywords

  • Precision livestock farming
  • Parturition
  • Bovine
  • Random forest
  • Animal-mounted sensors

Cite this

Miller, G A ; Mitchell, MA ; Barker, ZE ; Giebel, Katharina ; Codling, Edward ; Amory, Jonathan ; Michie, C ; Davison, Chris ; Tachtatzis, Christos ; Andonovic, Ivan ; Duthie, C-A. / Using animal‐mounted sensor technology and machine learning to predict time-to-calving in beef and dairy cows. In: Animal. 2020.
@article{4b3f11d610da47fdad2faf0983398cdc,
title = "Using animal‐mounted sensor technology and machine learning to predict time-to-calving in beef and dairy cows",
abstract = "Worldwide, there is a trend towards increased herd sizes and the animal to stockman ratio is increasing within the beef and dairy sectors, thus the time available to monitoring individual animals is reducing. The behaviour of cows is known to change in the hours prior to parturition, e.g. less time ruminating and eating, and increased activity level and tail raise events. These behaviours can be monitored non-invasively using animal mounted sensors. Thus behavioural traits are ideal variables for the prediction of calving. This study explored the potential of two sensor technologies for their capabilities in predicting when calf expulsion should be expected. Two trials were conducted at separate locations: i) beef cows (n = 144) and (ii) dairy cows (n = 110). Two sensors were deployed on each cow: 1) Afimilk Silent Herdsman (SHM) collars monitoring time spent ruminating (RUM), eating (EAT) and the relative activity level (ACT) of the cow and 2) tail mounted Axivity accelerometers to detect tail-raise events (TAIL). The exact time the calf was expelled from the cow was determined by viewing closed-circuit television camera footage. Machine learning random forest (RF) algorithms were developed to predict the when calf expulsion should be expected using single sensor variables and by integrating multiple sensor data-streams. The performance of the models were tested by the Matthew’s Correlation Coefficient (MCC), the area under the curve and the sensitivity and specificity of predictions. The TAIL model was slightly better at predicting calving within a five hour window for beef cows (MCC = 0.31) than for dairy cows (MCC = 0.29). The TAIL+RUM+EAT models were equally as good at predicting calving within a five hour window for beef and dairy cows (MCC = 0.32 for both models). Combining data-streams from SHM and tail sensors did not substantially improve model performance over tail sensors alone therefor hour-by-hour algorithms for the prediction of the time of calf expulsion were developed using tail sensor data. Optimal classification occurred at two hours prior to calving for both beef (MCC = 0.29) and dairy cows (MCC = 0.25). This study has shown that tail sensors alone are adequate for the prediction of parturition and that the optimal time for prediction is two hours before expulsion of the calf.",
keywords = "Precision livestock farming, Parturition, Bovine, Random forest, Animal-mounted sensors",
author = "Miller, {G A} and MA Mitchell and ZE Barker and Katharina Giebel and Edward Codling and Jonathan Amory and C Michie and Chris Davison and Christos Tachtatzis and Ivan Andonovic and C-A Duthie",
year = "2020",
month = "1",
day = "13",
doi = "10.1017/S1751731119003380",
language = "English",
journal = "Animal",
issn = "1751-7311",
publisher = "Cambridge University Press",

}

Using animal‐mounted sensor technology and machine learning to predict time-to-calving in beef and dairy cows. / Miller, G A; Mitchell, MA; Barker, ZE; Giebel, Katharina ; Codling, Edward; Amory, Jonathan; Michie, C; Davison, Chris; Tachtatzis, Christos; Andonovic, Ivan; Duthie, C-A.

In: Animal, 13.01.2020.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Using animal‐mounted sensor technology and machine learning to predict time-to-calving in beef and dairy cows

AU - Miller, G A

AU - Mitchell, MA

AU - Barker, ZE

AU - Giebel, Katharina

AU - Codling, Edward

AU - Amory, Jonathan

AU - Michie, C

AU - Davison, Chris

AU - Tachtatzis, Christos

AU - Andonovic, Ivan

AU - Duthie, C-A

PY - 2020/1/13

Y1 - 2020/1/13

N2 - Worldwide, there is a trend towards increased herd sizes and the animal to stockman ratio is increasing within the beef and dairy sectors, thus the time available to monitoring individual animals is reducing. The behaviour of cows is known to change in the hours prior to parturition, e.g. less time ruminating and eating, and increased activity level and tail raise events. These behaviours can be monitored non-invasively using animal mounted sensors. Thus behavioural traits are ideal variables for the prediction of calving. This study explored the potential of two sensor technologies for their capabilities in predicting when calf expulsion should be expected. Two trials were conducted at separate locations: i) beef cows (n = 144) and (ii) dairy cows (n = 110). Two sensors were deployed on each cow: 1) Afimilk Silent Herdsman (SHM) collars monitoring time spent ruminating (RUM), eating (EAT) and the relative activity level (ACT) of the cow and 2) tail mounted Axivity accelerometers to detect tail-raise events (TAIL). The exact time the calf was expelled from the cow was determined by viewing closed-circuit television camera footage. Machine learning random forest (RF) algorithms were developed to predict the when calf expulsion should be expected using single sensor variables and by integrating multiple sensor data-streams. The performance of the models were tested by the Matthew’s Correlation Coefficient (MCC), the area under the curve and the sensitivity and specificity of predictions. The TAIL model was slightly better at predicting calving within a five hour window for beef cows (MCC = 0.31) than for dairy cows (MCC = 0.29). The TAIL+RUM+EAT models were equally as good at predicting calving within a five hour window for beef and dairy cows (MCC = 0.32 for both models). Combining data-streams from SHM and tail sensors did not substantially improve model performance over tail sensors alone therefor hour-by-hour algorithms for the prediction of the time of calf expulsion were developed using tail sensor data. Optimal classification occurred at two hours prior to calving for both beef (MCC = 0.29) and dairy cows (MCC = 0.25). This study has shown that tail sensors alone are adequate for the prediction of parturition and that the optimal time for prediction is two hours before expulsion of the calf.

AB - Worldwide, there is a trend towards increased herd sizes and the animal to stockman ratio is increasing within the beef and dairy sectors, thus the time available to monitoring individual animals is reducing. The behaviour of cows is known to change in the hours prior to parturition, e.g. less time ruminating and eating, and increased activity level and tail raise events. These behaviours can be monitored non-invasively using animal mounted sensors. Thus behavioural traits are ideal variables for the prediction of calving. This study explored the potential of two sensor technologies for their capabilities in predicting when calf expulsion should be expected. Two trials were conducted at separate locations: i) beef cows (n = 144) and (ii) dairy cows (n = 110). Two sensors were deployed on each cow: 1) Afimilk Silent Herdsman (SHM) collars monitoring time spent ruminating (RUM), eating (EAT) and the relative activity level (ACT) of the cow and 2) tail mounted Axivity accelerometers to detect tail-raise events (TAIL). The exact time the calf was expelled from the cow was determined by viewing closed-circuit television camera footage. Machine learning random forest (RF) algorithms were developed to predict the when calf expulsion should be expected using single sensor variables and by integrating multiple sensor data-streams. The performance of the models were tested by the Matthew’s Correlation Coefficient (MCC), the area under the curve and the sensitivity and specificity of predictions. The TAIL model was slightly better at predicting calving within a five hour window for beef cows (MCC = 0.31) than for dairy cows (MCC = 0.29). The TAIL+RUM+EAT models were equally as good at predicting calving within a five hour window for beef and dairy cows (MCC = 0.32 for both models). Combining data-streams from SHM and tail sensors did not substantially improve model performance over tail sensors alone therefor hour-by-hour algorithms for the prediction of the time of calf expulsion were developed using tail sensor data. Optimal classification occurred at two hours prior to calving for both beef (MCC = 0.29) and dairy cows (MCC = 0.25). This study has shown that tail sensors alone are adequate for the prediction of parturition and that the optimal time for prediction is two hours before expulsion of the calf.

KW - Precision livestock farming

KW - Parturition

KW - Bovine

KW - Random forest

KW - Animal-mounted sensors

U2 - 10.1017/S1751731119003380

DO - 10.1017/S1751731119003380

M3 - Article

C2 - 31928536

JO - Animal

JF - Animal

SN - 1751-7311

ER -