Validation of RumiWatch nose-band sensors for measuring nutritional behaviour of dairy cows
Year: 2016
Type: dissertation
Abstract: Changes in rumination and feeding behaviour may be signs of health problems in cattle.Automated systems for monitoring behaviour can be an effective tool in the prevention and diagnosis of the disease in cattle in the early stages.The objective of the thesis was to evaluate the reliability of the functioning of the RumiWatch System (RWS)a sensor-based device that measures ruminating, feeding and drinking behaviour in dairy cattle.The device registers the cow's jaw movements through a pressure sensor.RumiWatch noseband sensors were attached to five non-lactating dairy cows of the breeds Nordic Red and Holstein.Video recording observation was used to validate the system by using a confusion matrix method.In terms of the performance indicators, rumination behaviour accuracy was 89%, and RWS was especially good in differentiating "not ruminating" from "ruminating" (specificity 93%).However, it was slightly worse in recognizing all true rumination cases (sensitivity 78%), and classified some of the other behaviours erroneously as "rumination" (precision 79%).For eating behaviour, accuracy was much lower (67%) than for ruminating, resulting mainly from poor precision (51%) and to some extent from lower specificity (82%).Instead, RWS's sensitivity was slightly higher for eating (82%) than ruminating.In the case of drinking, accuracy (98%) and specificity (99%) were very high but precision (6%) and sensitivity (7%) were extremely low, which reflected the accuracy paradox of the imbalanced data for drinking.In conclusion, the RWS in this study proved to be a relatively useful device for measuring ruminating and eating but not drinking.RWS might become a valuable tool for researchers and farmers in the future, but further validation is desirable.The confusion matrix approach is useful in the validation of RWS, as well as other devices, since it reveals the types of the misclassifications a classifier makes, which helps to adjust the classifying algorithm.УНИВЕРСИТЕТ ВОСТОЧНОЙ ФИНЛЯНДИИ Факультет естественных наук и лесного хозяйства, Кафедра Окружающей среды и Биологических наук Зеленая Биотехнология и Пищевая безопасность Динара Кабылбекова: Проверка аппарата "RumiWatch" для измерения пищевого поведения дойных коров Магистерская работа, 38 страниц Руководители: Профессор Яко Мононен и MSc Салла Рууска Сентябрь 4, 2016 _____________________________________________________________________________ Ключевые слова: дойные коровы
Author: Dinara Kabylbekova
Language: English
Subfield: Animal Science and Zoology
Domain: Life Sciences
SDG: Zero hunger
Open Access status: green
OpenAlex ID: W3095011961