Development of calving prediction model using sensor technologies in dairy cows

The aim of the presented thesis study is to detect physiological and behavioral changes observed during the prepartum period in Holstein dairy cows using sensor technologies and develop calving time prediction models based on this data. For this purpose, multiparous Holstein cows (n=28) were used to examine the prepartum period of 1-2 weeks. The pregnant cows were fitted with a vajinal device containing temperature, pressure, conductivity, and accelerometer sensors, as well as an ear tag with an accelerometer sensor for rumination measurement. The intravajinal device was inserted into the anterior vajina with the help of an auxiliary apparatus after the vulva area was disinfected with a 10% povidone iodine solution. The vajinal device sent data every 2.33 minutes, while the rumination sensor sent data every 20 minutes. The obtained data were stored and archived in a database system. Animal behaviors were observed in real-time monitoring system, and after all calvings, ease of calving, stillbirth scoring, and calf vigor scoring were performed. The rupture of the chorioallantoic membrane and regular abdominal contractions during the onset of the second stage of labor facilitated the removal of the device from the vajina. The time of device removal was determined by detecting the drop in temperature recorded in the JSP file, and the calving time was determined by analyzing instant camera recordings. When the processed data of the 1-2 week prepartum period were examined, it was determined that the hourly average vajinal temperature showed compliance with a daily diurnal rhythm between 39.0-39.2 °C, except for the last 20-24 hours. Based on measurements taken from the vajinal environment using the activity sensor (G sensor), it was observed that the hourly total activity data showed a scattered structure, varying degrees of increase/decrease throughout the day, and did not exhibit a specific change close to calving. When the daily average activity data were examined, a slight increase in activity was observed 24 hours before calving. The daily average conductivity values showed a similar trend to the activity data. A moderate decrease in the daily average conductivity values was determined 24 hours before calving. It was determined that there was an average increase of 20-30 mbar in the pressure data during lying down by matching the data from the pressure sensor with the camera images, and this change enabled the differentiation of lying down and standing up behaviors. According to the pressure data, there was a gradual increase in the daily average standing and lying down frequencies over a 4-day prepartum period, while there was a gradual decrease in the average lying time and the average time spent during a lying period. In conclusion, it is possible to facilitate economically valuable routine practices such as calving using sensor technologies. It is believed that the use of combined systems instead of single sensor usage will increase the success rate.