Advancements in the analysis of behavioural and performance data for early lameness detection in dairy cows

Introduction
According to a study by the European Food Safety Association, despite the technological and scientific advancements in the field of dairy farming, there has been no significant reduction of lameness prevalence on dairy farms in the last decades [1]. Claw lesions and diseases are a cause of pain and suffering for the affected animals, making lameness an important animal welfare issue. Moreover, lameness causes economic losses due to expenditures originating from treatment costs as well as reduced performance. Lame cows have a reduced milk yield [2], [3], are affected in their reproductive efficiency and have an increased risk of culling and death [4]. Early lameness detection is crucial in order to avoid both economic losses related to lameness and pain for the animals [5], but due to the structural changes taking place in the dairy farms individual animal observation is either not possible or not carried out often enough. Exhibiting of pain or distress exposes cows, making them vulnerable to predators. For this reason, both pain recognition and locomotion scoring are not ideal methods for recognising lame animals at an early stage of the disease. Farmers in fact tend to underestimate [6] the lameness prevalence on their farms and to treat lameness at an advanced stage [5], [7]. Automatic lameness detection would help farmers recognise and treat lameness at an earlier stage. The aim of this study was to test and further develop a predictive model developed in a previous project at the Institute for Agricultural Engineering and Animal Husbandry [8] on commercial dairy farms in Lower Bavaria. […]