Ovulation-Anchored Evaluation of IMU-Derived Activity and Posture-Related Behavioral Changes Across Natural Estrus Phases in Dairy Cattle

Accurate estrus detection is essential for optimizing artificial insemination timing, but visual detection is limited by labor demands, intermittent observation, short estrus duration, and variable behavioral expression. Although inertial measurement unit (IMU) systems capture dynamic acceleration and rotational movement, phase-specific IMU-derived activity and posture-related changes during natural estrus remain insufficiently characterized. Therefore, this study evaluated these variables across an ovulation-anchored six-phase framework using video-derived behavioral observations and ultrasound-confirmed ovulation as biological reference standards. In this observational study, five dairy cows contributing eleven natural estrus cycles were monitored, yielding 285,337 time-aligned 10 s sensor observations that were summarized for phase-level analysis. Cow movement was recorded at 10 s intervals using neck-mounted tri-axial accelerometers and gyroscopes, while posture states, estrus-related behaviors, and ovulation timing were determined from continuous video recordings and 6 h transrectal ultrasonography. Extracted variables included signal vector magnitude, VeDBA, Gyro_mag, baseline-adjusted activity features, exploratory Combined Activity Index, posture proportions, and lying bout characteristics. VeDBA was highest during standing estrus, whereas Gyro_mag and the Combined Activity Index increased during pre-estrus and standing estrus. Standing estrus involved less lying, more standing and walking, and shorter lying bout duration. These findings identify candidate IMU-derived and posture-related variables for future standing-estrus differentiation models and potential insemination-timing support, pending validation in larger independent populations.