Management practices and the use of automated technologies to improve calf health and welfare on dairy farms

Abstract

The objective of the research reported in this dissertation was to investigate management practices and the use of automated technologies to improve calf health and welfare on dairy farms. Three independent studies were conducted to address the main goals of this dissertation, with the following specific objectives, (1) to investigate the association between feeding behaviors and management practices, (2) to investigate the association between feeding behaviors and disease, and (3) to create a predictive model for disease detection using feeding behaviors in automatically fed group-housed preweaned dairy calves in the Upper Midwest U.S. In order to assess the associations between feeding behaviors and management factors, personnel visited 25 farms (including a total 2,413 calves) on a bimonthly basis over an 18-mo period to collect automated milk feeder (AMF) software feeding behavior data and various management factors of interest. Linear regressions using Pearson’s correlation were fit with feeding behaviors (drinking speed, unrewarded visits, rewarded visits, and consumption percentage) as the response variables and management factors (light intensity at the feeder (LIF) and bedding depth (BDP), amount of milk at peak (MAP), age of calves at weaning (WEA), number of calves housed in the group (NUM), and the average fat percent in milk (AOF)) as the predictors. The LIF, BDP, NUM, MAP, WEA, and AOF were associated with feeding behaviors (P < 0.05). This exploratory study provided novel information that management practices might influence preweaned dairy calf feeding behaviors. In order to assess the associations between feeding behaviors and disease, personnel visited a singular farm (included a final dataset of 599 Holstein heifer calves) on a weekly basis over a 1-yr period to collect AMF feeding behavior data (milk intake (mL/d), drinking speed (mL/min), visit duration (min), rewarded (with milk being offered) and unrewarded (without milk) visits (number per d), and interval between visits (min)) and visually health score calves. Calf health scores included calf attitude, ear position, ocular discharge, nasal discharge, hide dirtiness, cough score, and rectal temperatures. Generalized additive mixed models (GAMM) were used to identify associations between feeding behavior and disease. Total milk intake (mL/d), drinking speed (mL/min), interval between visits (min) to the AMF, calf age (d), and rewarded visits were significantly associated with dairy calf health status (P < 0.05). This study suggests that AMF data may be a useful screening tool for detecting disease in dairy calves. Finally, in order to create a predictive model for disease detection using feeding behaviors from AMF, personnel visited a singular farm on a weekly basis over a 1-yr period to collect AMF feeding behavior data, collect calves’ treatment records, and visually health score Holstein heifer calves for attitude, ear position, ocular discharge, nasal discharge, hide dirtiness, and cough score. The final data sets consisted of 719 and 741 calves with 1,594 and 1,044 observations for Data 1 (visual health scores to predict disease) and Data 2 (treatment records to predict disease), respectively. Calf visit-level feeding behaviors from AMF data included milk intake (mL/d), drinking speed (mL/min), visit duration (min), rewarded (with milk being offered) and unrewarded (without milk) visits (number per d), and interval between visits (min) were used to predict health status with 16 machine learning algorithms per machine learning approach (e.g., Generalized Linear Model, Random Forest, Gradient Boosting Machine). This study suggests that machine learning was effective in determining specific visit-level feeding behaviors to predict disease in group-housed preweaned dairy calves.