![](/upload_mm/8/c/2/1c13ed97-dd53-4179-9eb1-d8b8f7b4ffa0_shutterstock_1399837376_1142a506_750x400.jpg)
Project
Quantification of climate impact for breeding and decision support at dairy farms
We develop machine learning models that predict the sensitivity of individual cows to heat stress based on their behaviour. Behavioural sensor data, cow information, and climatic data were collected from farms in the Netherlands and Belgium. By integrating multiple data sources, the models could better support dairy farms' adaptation to climate change. Farmers can utilise this information to differentiate between cows that are susceptible to heat stress and those that can tolerate it, which supports farm management and breeding decisions.
Climate change results in an increased frequency of extreme weather conditions. Dairy cows have a high metabolic rate and, therefore, they easily suffer from heat stress. Heat stress decreases appetite, affects rumen health, and leads to higher disease occurrence. Overall, these effects have negative consequences on milk production, fertility and wellbeing. Previously, we investigated the impact of heat stress on group-level and individual cow production and behaviour using a method that allowed to distinguish the group-level and individual differences, and quantify short and long-term effects. Our research showed that individual cows differ in their heat stress sensitivity, both in terms of production and behavioural responses. Building upon our previous work, in this project we aim to develop a machine learning model that predicts the severity of heat stress impact in dairy cows using data from behavioural sensors.
Project description
High-frequency behavioural sensor data were collected from multiple dairy farms in the Netherlands and Belgium. For all farms, cow information and climatic data are also available. Using the individual cow behavioural time series and the relevant covariates (e.g. parity), we first construct biologically meaningful features by combining domain knowledge with high-frequency behavioural sensor data. We use these features to develop machine learning models that predict the individual cows’ sensitivity to heat stress. We apply both univariate (impact on production and health separately) and multivariate approaches (production and health combined). With the integration of multiple data sources, we further improve the current use of these data streams to support the adaptation of dairy farms to the changing climate. Upon better understanding and prediction of the impact of heat stress on dairy cows in different farm environments, farmers and breeders can (1) better anticipate the impact of the stressor for individual cows, and make decisions that reduce its short and long-term impact on production, wellbeing and health; and (2) select the best animals for specific environments by identifying which cows deal better with heat stress.