Experienced veterinary surgeons require multiple attempts but gain proficiency rapidly in robotic simulator tasks.

OBJECTIVE: To assess attempts to proficiency of experienced veterinary surgeons for 2 surgical tasks when using a robotic simulator (Mimic dV-Trainer; Surgical Sciences) and determine factors associated with the successful performance of these tasks. METHODS: Veterinary surgeons with rigid, minimally invasive surgery experience performed 2 tasks ("pick and place" and "knot the ring 1") using the simulator until they attained proficiency. Individual performance variables were recorded. The number of attempts to proficiency was recorded. Performance variables were also assessed for effect on proficiency by the Kendall tau correlation and hierarchical multiple linear regression. The study period was from July 25, 2022, through December 14, 2022. RESULTS: The 18 surgeons enrolled required a median of 8.5 attempts (95% CI, 7 to 12; range, 6 to 22) to reach proficiency for the basic task versus 27 attempts (95% CI, 21 to 38; range, 10 to 63) for the advanced task. Surgeons took a median of 6 minutes (range, 3 to 11 minutes) to complete training for the basic task and 12 minutes (range, 4 to 46 minutes) for the advanced task. The number of attempts to reach proficiency correlated strongly with economy of motion (tau = 0.72), instrument collisions (tau = 0.72), and time to completion (tau = 0.96). CLINICAL RELEVANCE: Although experienced surgeons required a high number of attempts to gain proficiency in robotic simulator tasks, they did achieve proficiency quickly, encouraging future investigations into their use for training. Specific motion metrics were identified which improved efficiency during training.

Looking to the Future; Veterinary Robotic Surgery.

This article details the rise of surgical robots in the human surgical sphere as well as their use in veterinary medicine. Sections will describe in detail the equipment required for these procedures and the advantages and disadvantages of their use. Specific attention is given to the articulated instrumentation, which affords psychomotor benefits not only for surgical precision but also for surgeon ergonomics. A discussion of the possible indications and current use of robotics in veterinary medicine and the challenges to integrating robotics is also provided.

Association between management practices and estimated mastitis incidence and milk losses on robotic dairy farms.

This study aims to describe the relation between farm-level management factors and estimated farm-level mastitis incidence and milk loss traits (MIMLT) at dairy farms with automated milking systems. In this observational study, 43 commercial dairy farms in Belgium and the Netherlands were included and 148 'management and udder health related variables' were obtained during a farm visit through a farm audit and survey. The MIMLT were estimated from milk yield data. Quarter-level milk yield perturbations that were caused by presumable mastitis cases (PMC) were selected based on quarter-level milk yield and electrical conductivity. On average, 57.6 ± 5.4% of the identified milk yield perturbations complied with our criteria. From these PMC, 3 farm-level MIMLT were calculated over a one-year period around the farm visit date: (1) the 'average number of PMC per cow per year', (2) the 'absolute milk loss per cow per day', calculated as the farm-level sum of all milk losses during PMC in one year, divided by the average number of lactating cows and the number of days, and (3) the 'relative milk loss', calculated as the farm-level sum of milk losses during PMC in one year, divided by the estimated total production in the absence of PMC. The 'average number of PMC per cow per year' was on average 1.81 ± 0.47. The PMC caused an average milk loss of 0.77 ± 0.26 kg per lactating cow per day, which corresponded to an average production loss of 2.38 ± 0.82% of the expected production in the absence of PMC. We performed a principal component regression (PCR) analysis to link the 3 MIMLT to the 'management and udder health related variables', whilst reducing the multicollinearity and the number of dimensions. The first principal component was mainly related to 'milking system brand, maintenance and settings'. The second component mainly linked to average productivity and somatic cell counts, whereas the third component mainly contained variables linked with mastitis management, treatment, and biosecurity. The 3 PCR models had R² ranging from 0.46 (for absolute milk loss per cow per day) to 0.57 (for relative milk loss). For all models, the second PC had the largest effect size. This analysis raises awareness of the impact of management factors on a factual basis and provides handles to take management actions to improve udder health.

Genetic and phenotypic variation and consistency in cow preference and circadian use of robotic milking units.

Many dairy herds use automatic milking stations (AMS), with cows in large herds often having access to 2 or more AMS, and must choose between them when they go for milking. Individual cows acquire routines of either consistently using a specific milking box or consistently using any available milking box. Here, we hypothesized that the degree of use of the same milking box was an expression of preference, and quantified it as preference consistency score (PCS). The PCS was calculated as a ratio between the excess frequencies of the first choice over the base frequency of "not first choice" over 15-d segments of lactation. This ratio was 0 if all choices were taken equally, and became 1.0 if only the first choice was taken in all events. We investigated the consistency of milking box preference in 2 cohorts (one Holstein and one Jersey) across 6 commercial dairy herds in Denmark (n = 4,665 cows total). In addition to PCS, we recorded and analyzed associated milking and behavior traits, including a time profile index showing use of specific clock hours when cows were milked (Time_profile, based on excess use of specific clock hours), milking frequency, time spent in the milking box, and milk yield. Records from each milking event were condensed into 15-d segments based on days in milk. The data were analyzed using a linear mixed model, with random genetic and individual cow effects, to estimate heritability (h2), repeatability (t), and individual level correlations (ri) between traits. The average PCS was 0.43 and 0.41 in Holstein and Jersey, respectively, showing that cows developed routines for consistently using the same milking box; however, some cows had lower preference (i.e., greater flexibility in use). The Time_profile indicated that some cows were milked in a few hour-bins, whereas others were more flexible. The PCS and Time_profile traits had low heritability (h2, PCS/Time_profile = 0.07 ± 0.02/0.11 ± 0.02 Holstein, 0.13 ± 0.03/0.04 ± 0.02 Jersey) and moderate repeatability (t, PCS/Time_profile = 0.47/0.40 Holstein, 0.50/0.42 Jersey). The 2 traits were weakly correlated with each other (ri = 0.18 and 0.17), and were weakly correlated with milk yield (ri range: 0.0 to -0.10). However, the time profile was strongly correlated with milking frequency (ri range: -0.81 to -0.73), and was moderately correlated with daily box time (ri range: -0.43 to -0.35). In general, Holstein and Jersey parameter estimates were of similar size, and thus in good agreement. Overall, individual cows covered a broad spectrum of preference consistency, both regarding the use of specific milking boxes and time profiles, with these 2 traits representing different aspects or dimensions of milking behavior. The findings that some cows have strong preferences for specific AMS may be most useful in herd management and farm design. The weak correlation to milk yield indicated that yield minimally affected these 2 milking associated behavior traits. In conclusion, although the traits were repeatable, heritability was low; thus, genetic selection for milk yield might minimally affect these 2 traits.