1. Lameness Detection in Sheep Through the Analysis of the Wireless Sensor Data The School of Science and Technology Department of Computer Science and Immersive Technologies Zainab Al-Rubaye Supervisory Team 1 st Supervisor: Dr. Ali Al-Sherbaz 2 nd Supervisor: Dr. Wanda McCormick Director of the Study: Dr. Scott Turner 14/06/2016

2. Outlines: Problem statement. Research aims. Data sensor examples. Raw data collection. Research Methodology. Implementation/ what next … 14/06/2016

3. Problem statement Lameness in sheep is a clinical manifestation of painful disorders, mainly related to the locomotor system, resulting in impaired movement or deviation from normal gait or posture (Van Nuffel, Zwertvaegher, Pluym, et al. 2015). Lameness is a common cause of welfare and economic concerns in most sheep-keeping countries. (UK rainy weather result in a humid soil). Annual lost from footrot alone is estimated by £6 per ewe in Great Britain Agriculture and Horticulture Development Board (AHDB). (Reduction lameness for better returns, manual7 2014). It varies from mildly lame to severely lame. 14/06/2016

4. Research Aims: To develop an automated model to early detect lameness in sheep by analysing the data that will be retrieved from a mounted sensor on the sheep neck. This model will help the shepherd to early detect the lame sheep to prevent the worse situation of trimming or even culling the sheep. 14/06/2016

5. Time Accuracy )one reading every 40 mS( -25 readings/ S (1000/40) -25*60=1500 reading/Min. -1500*60= 90000 / Hr. Acceleration M/S 2 X Accelerometer Y Accelerometer Z Accelerometer Angular velocity (Rad/s) X Angular velocity Y Angular velocity Z Angular velocity Orientation (clockwise/anticlockwise) Roll angle (Deg around X axis) Pitch angle (Deg around Y axis) Head angle (Deg around Z axis) Longitude latitude Data sensor example y 14/06/2016

6. Mobile Application as a prototype sensor: 14/06/2016

7. Row data collection example1 (Lame sheep): 14/06/2016

8. Row data collection example2 (Sound sheep): 14/06/2016

9. Observe for period of time. and data are collected Research Methodology Classification (decision tree) Validate with test data set Algorithm Enhancement Divide into training set and test set Sensor device 14/06/2016

10. Implementation: The Sensor gives the lameness alarm The developed Algorithm will built in the sensor itself. The Base Station gives the lameness alarm The develop algorithm will be in a remote base station (communication part may be need) 14/06/2016

11. What to do next … The proposed Decision/Regression tree Algorithm will be used due to its fast computational time. Readjust the sensor setting to give less readings per second. Eliminate the variable data sensor that have less effect on making a decision (identify lameness class). Reapply the algorithm until reach an optimal solutions. 14/06/2016

12. References: AHDB Beef & Lamb, Agriculture & Horticulture Development Board. 2014. Manual 7 - Reducing lameness for better returns. [ONLINE] Available at: http://beefandlamb.ahdb.org.uk/returns/health-and-fertility/. [Accessed 01 March 16]. http://beefandlamb.ahdb.org.uk/returns/health-and-fertility/ Lynn Greiner. 2011. DataBase Trends and Applications. [ONLINE] Available at: http://www.dbta.com/Editorial/Trends-and-Applications/What-is-Data- Analysis-and-Data-Mining-73503.aspx. [Accessed 01 March 16].http://www.dbta.com/Editorial/Trends-and-Applications/What-is-Data- Analysis-and-Data-Mining-73503.aspx Tan, P.N., Steinbach, M. and Kumar, V., 2006. Introduction to data mining (Vol. 1). Boston: Pearson Addison Wesley. Vázquez Diosdado, J.A. et al., 2015. Classification of behaviour in housed dairy cows using an accelerometer-based activity monitoring system. Animal Biotelemetry, 3(1), p.15. Available at: http://www.animalbiotelemetry.com/content/3/1/15. Van Nuffel, A., Zwertvaegher, I., Pluym, L., et al., 2015. Lameness Detection in Dairy Cows: Part 1. How to Distinguish between Non-Lame and Lame Cows Based on Differences in Locomotion or Behavior. Animals, 5(3), pp.838–860. Available at: http://www.mdpi.com/2076-2615/5/3/0387/.http://www.mdpi.com/2076-2615/5/3/0387/ 14/06/2016

13. Any Questions are Welcomed ….. 14/06/2016

14. Data Analysis: Data mining is the process of automatically retrieving useful information from massive data repository by predicting the results of future observations Data mining is a confluence of many disciplines and this interdisciplinary feature contributes significantly to implement very successful data mining applications and development research. Data Mining Statistical Techniques Machine Learning Techniques 14/06/2016

15. Relevance work approaches : 14/06/2016 Data Collection Methods Human Observation Video Cameras Sensor data (GPS, Accelerometer, head movements,....) Data Analysis Methods LS /GS Scoring system techniques/done by trained observer) Statistical Techniques Computerized techniques (Data mining use Machine Learning techniques) Analysis Purpose Detect animal illness (mastitis, lameness, ketosis) Classification (lying, standing, grazing, ruminating) Species Type Cattle sheep Other species