Using successful inseminations to evaluate sensitivity of automated heat detection systems

Automated oestrus detection systems are becoming a mainstream technology on dairy farms. A previous study demonstrated that these systems generate heat alerts too early or too late in comparison to progesterone-defined oestrus events. That study, however, included only 30 cases. The current study used successful inseminations as reference to evaluate sensitivity of heat detection systems. Data were collected on two farms from January through July 2014 and included insemination records and heat alerts from three different automated heat detection system. One system was implemented on both farms, and each farm had a second, different, heat detection system running simultaneously. Insemination records were used to identify successful inseminations (n = 282; 153 on Farm 1 and 129 on Farm 2). The day each system generated heat alerts was compared with the day of successful insemination to calculate the sensitivity for successful inseminations. Sensitivity was also assessed per parity (1, 2 or ≥3) and per lactation stage (≤56 and >56 days in milk) within each system. Using a two-day time window (day of successful insemination and previous day) for a system to generate an alert, sensitivities were 35%, 87%, and 94% for system A, B, and C, respectively. Systems had lower sensitivity levels for heifers (parity 1) and system B appeared to have higher sensitivity levels for successful inseminations during the first 56 days of lactation. Despite small differences in sensitivity levels for the different parity groups and lactation stages, there appeared to be no differences of practical relevance. Results suggested no necessity for parity specific or lactation stage specific algorithms to improve detection performance of heat detection systems.