Predicting Lameness in Sheep Activity Using Tri-Axial Acceleration Signals

Abstract

Simple SummaryMonitoring livestock farmed under extensive conditions is challenging and this is particularly difficult when observing animal behaviour at an individual level. Lameness is a disease symptom that has traditionally relied on visual inspection to detect those animals with an abnormal walking pattern. More recently, accelerometer sensors have been used in other livestock industries to detect lame animals. These devices are able to record changes in activity intensity, allowing us to differentiate between a grazing, walking, and resting animal. Using these on-animal sensors, grazing, standing, walking, and lame walking were accurately detected from an ear attached sensor. With further development, this classification algorithm could be linked with an automatic livestock monitoring system to provide real time information on individual health status, something that is practically not possible under current extensive livestock production systems.AbstractLameness is a clinical symptom associated with a number of sheep diseases around the world, having adverse effects on weight gain, fertility, and lamb birth weight, and increasing the risk of secondary diseases. Current methods to identify lame animals rely on labour intensive visual inspection. The aim of this current study was to determine the ability of a collar, leg, and ear attached tri-axial accelerometer to discriminate between sound and lame gait movement in sheep. Data were separated into 10 s mutually exclusive behaviour epochs and subjected to Quadratic Discriminant Analysis (QDA). Initial analysis showed the high misclassification of lame grazing events with sound grazing and standing from all deployment modes. The final classification model, which included lame walking and all sound activity classes, yielded a prediction accuracy for lame locomotion of 82%, 35%, and 87% for the ear, collar, and leg deployments, respectively. Misclassification of sound walking with lame walking within the leg accelerometer dataset highlights the superiority of an ear mode of attachment for the classification of lame gait characteristics based on time series accelerometer data.

Identifying Sheep Activity from Tri-Axial Acceleration Signals Using a Moving Window Classification Model

“…To overcome this, up or under-sampling to equalize the data for each behaviour category in the training dataset can be used. Sakai et al [11] compared the precision and sensitivity of behaviour classification before and after balancing, reporting mixed results. A similar approach may be warranted with the current study to determine the effect of balancing data on the classification performance of the testing dataset.…”

“…Such information could be used to create animal health and welfare indicators on the basis of deviations in activity patterns from baseline levels [2,3]. The capability of accelerometers to measure posture and activity states has been well established in ruminants (cattle [4][5][6][7][8], goats [9][10][11] and sheep [3,[12][13][14][15][16][17]).…”

Invited review: Cattle lameness detection with accelerometers

“…Individual steps from normal and block attached walking were then classified by the model as normal or anomalous and reported an accuracy of 91%. A similar approach to simulating lameness has been reported in sheep where one leg was taped in such a way to restrict locomotion (Barwick et al, 2018). However, it has not been tested if these are valid proxies for the abnormal locomotion typically associated with lameness.…”

“…Their classifiers were built using decision trees, additive logistic regression and logit boost. Other approaches that could be or have been applied include K-nearest neighbors and random forest (Byabazaire et al, 2019), time series analysis (Maertens et al, 2011;Barwick et al, 2018), wavelet analysis (Pastell et al, 2009), and a range of machine learning approaches (Valletta et al, 2017). In summary, logistic regression and support vector machines are the 2 most prominent approaches.…”

Automated detection of lameness in sheep using machine learning approaches: novel insights into behavioural differences among lame and non-lame sheep

“…In comparison to cattle, there is limited work in sheep [10][11][12], with almost all studies focused on only classifying basic activities, such as standing, grazing and lying, without identifying different features that classify lameness. A recent study by Barwick et al [12] using five sheep investigated the classification of lame walking activity in sheep; however, as mentioned above, walking only constitutes a very limited part of daily activity budget of sheep which limits its overall utility and does not give insight into the whole behaviour of a lame sheep.…”