TY - JOUR
T1 - Delineating Mastitis Cases in Dairy Cows
T2 - Development of an IoT-Enabled Intelligent Decision Support System for Dairy Farms
AU - Khan, Mohammad Farhan
AU - Thorup, Vivi Morkore
AU - Luo, Zhenhua
PY - 2024/7
Y1 - 2024/7
N2 - —Mastitis, an intramammary bacterial infection, is not only known to adversely affect the health of a dairy cow but also to cause significant economic loss to the dairy industry. The severity and spread of mastitis can be restrained by identifying the early signs of infection in the cows through an intelligent decision support system. Early intervention and control of infection largely depend on the availability of on-site high throughput machinery, which can analyze milk samples regularly. However, due to limited resources, marginal and small farms usually cannot afford such high-end machinery, hence, the financial loss in such farms due to mastitis may become significant. To overcome such limitations, this article proposes a low-complexity yet affordable automated system for accurate prediction of early signs of clinical mastitis infection in dairy cows. In this work, behavioral data collected through Internet of Things (IoT)-enabled wearable sensors for cows is utilized to develop a support vector machine (SVM) model for the daily prediction of mastitis cases in a dairy farm. The dataset from the research herd utilizes the information of 415 cows collected in the span of 4.75 years in which 75 cows had mastitis. In addition to relevant behavioral features, other statistically significant features, such as daily milk yield, lactation period, and age are also utilized as features. Our study indicates that the SVM model comprising a subset of behavioral and nonbehavioral features can deliver a mastitis prediction accuracy of 89.2%.
AB - —Mastitis, an intramammary bacterial infection, is not only known to adversely affect the health of a dairy cow but also to cause significant economic loss to the dairy industry. The severity and spread of mastitis can be restrained by identifying the early signs of infection in the cows through an intelligent decision support system. Early intervention and control of infection largely depend on the availability of on-site high throughput machinery, which can analyze milk samples regularly. However, due to limited resources, marginal and small farms usually cannot afford such high-end machinery, hence, the financial loss in such farms due to mastitis may become significant. To overcome such limitations, this article proposes a low-complexity yet affordable automated system for accurate prediction of early signs of clinical mastitis infection in dairy cows. In this work, behavioral data collected through Internet of Things (IoT)-enabled wearable sensors for cows is utilized to develop a support vector machine (SVM) model for the daily prediction of mastitis cases in a dairy farm. The dataset from the research herd utilizes the information of 415 cows collected in the span of 4.75 years in which 75 cows had mastitis. In addition to relevant behavioral features, other statistically significant features, such as daily milk yield, lactation period, and age are also utilized as features. Our study indicates that the SVM model comprising a subset of behavioral and nonbehavioral features can deliver a mastitis prediction accuracy of 89.2%.
KW - Animal health informatics
KW - automated detection
KW - clinical mastitis
KW - Cows
KW - Dairy products
KW - decision support system
KW - Decision support systems
KW - Diseases
KW - Internet of Things (IoT) wearable sensor
KW - Legged locomotion
KW - Predictive models
KW - Support vector machines
UR - http://www.scopus.com/inward/record.url?scp=85190810784&partnerID=8YFLogxK
U2 - 10.1109/TII.2024.3384594
DO - 10.1109/TII.2024.3384594
M3 - Journal article
AN - SCOPUS:85190810784
SN - 1551-3203
VL - 20
SP - 9508
EP - 9517
JO - IEEE Transactions on Industrial Informatics
JF - IEEE Transactions on Industrial Informatics
IS - 7
ER -