Automated behavioral detection of mastitis in ewes

 

Performance Narrative:

The original objectives of this project were to: (1) develop algorithms to automatically monitor ewe behavior from data collected using novel wearable sensors; (2) examine changes in these behaviors associated with, and preceding, subclinical mastitis in a meat-breed flock; (3) develop an algorithm to identify ewes at risk of mastitis to ensure timely and appropriate treatment. In addition to these objectives, we have since added a fourth one: (4) evaluate the accuracy of the California Mastitis Test (CMT), somatic cell count (SCC), and thermal imaging of the udder as diagnostic tools for subclinical mastitis.

In total, 39 lactating ewes were enrolled in this experiment from March to December 2021. We collected milk samples from each udder half weekly until 8 to 10 weeks after lambing for bacteriologic, SCC and CMT tests. Thermal photos of the udder were also taken at the time of milk sampling. Video was recorded continuously throughout the lactation period. 24 of these ewes and their lambs were fitted with harnesses that held sensors that continuously recorded positional and accelerometer data throughout the lactation period.

We have completed data collection, and PhD student, Gretchen Peckler, is in the process of analyzing data and preparing 3 manuscripts for peer-reviewed publication. We are currently using the sensor and video data to develop algorithms that automatically classify ewe behavior according to features in the sensor data (Objective 1). We will then assess whether these behavioral classifications can predict mastitis (Objective 2), and if so, develop an algorithm that uses sensor data to identify the early

onset of this disease (Objective 3). Finally, we have finished extracting temperatures of the udder halves using thermal imaging software and will compare their diagnostic potential to CMT and SCC methods (Objective 4).

Estimate the Total Percentage (%) of work Completed on the Project…100%……. # Accomplishment/Activity Relevance to Objective
1 PhD student, Gretchen Peckler, presented preliminary results at a departmental poster competition (4th place winner) Results described the potential for using udder temperature to diagnose subclinical mastitis in ewes.
2 Descriptive analysis of mastitis prevalence and etiology Prevalence and etiology will inform diagnostic analyses.
3 Continuation of data analysis of positional and accelerometer data These data will be used to define relationship between behavior and subclinical mastitis.

onset of this disease (Objective 3). Finally, we have finished extracting temperatures of the udder halves using thermal imaging software and will compare their diagnostic potential to CMT and SCC methods (Objective 4).

Estimate the Total Percentage (%) of work Completed on the Project…100%……. # Accomplishment/Activity Relevance to Objective
1 PhD student, Gretchen Peckler, presented preliminary results at a departmental poster competition (4th place winner) Results described the potential for using udder temperature to diagnose subclinical mastitis in ewes.
2 Descriptive analysis of mastitis prevalence and etiology Prevalence and etiology will inform diagnostic analyses.
3 Continuation of data analysis of positional and accelerometer data These data will be used to define relationship between behavior and subclinical mastitis.
Outcome and Indicator Results to Date # Outcome/Indicator Quantifiable Results
1 Subclinical mastitis prevalence and etiology Of 1352 milk samples, 25% tested positive for mastitis. Major pathogens were coagulase-negative Staphylococcus species, Staph aureus, and Mannheimia Haemolytica. Prevalence increases with parity and litter size.
2 Relationship between udder temperature and mastitis Preliminary data indicate that infected udders are significantly hotter (35.5°C) than healthy udders (34.7°C).