Respiratory diseases commonly affect livestock species, negatively impacting animal’s productivity and welfare. The use of precision livestock farming (PLF) applied in respiratory disease detection has been developed for several species. The aim of this systematic review was to evaluate if PLF technologies can reliably monitor clinical signs or detect cases of respiratory diseases. A technology was considered reliable if high performance was achieved (sensitivity > 90% and specificity or precision > 90%) under field conditions and using a reliable reference test. Risk of bias was assessed, and only technologies tested in studies with low risk of bias were considered reliable. From 23 studies included—swine (13), poultry (6), and bovine (4) —only three complied with our reliability criteria; however, two of these were considered to have a high risk of bias. Thus, only one swine technology fully fit our criteria. Future studies should include field tests and use previously validated reference tests to assess technology’s performance. In conclusion, relying completely on PLF for monitoring respiratory diseases is still a challenge, though several technologies are promising, having high performance in field tests.

Precision livestock farming technologies support health monitoring on farms, yet few studies have evaluated their effectiveness under field conditions using reliable gold standards. This study evaluated a commercially available technology for detecting cough sounds in pigs on a commercial farm. Audio was recorded over six days using 16 microphones across two pig barns. A total of 1110 cough sounds were labelled by an on-site observer using a cough induction methodology, and 8938 other sounds from farm recordings and open-source datasets (ESC-50, UrbanSound8K, and AudioSet) were labelled. A hybrid deep learning model combining Convolutional Neural Networks and Recurrent Neural Networks was trained and evaluated using these labels. A total of 34 audio features were extracted from 1 s segments, including validated descriptors (e.g., MFCC), unverified external features, and proprietary features. Features were evaluated through 10-fold cross-validation based on classification performance and runtime, resulting in eight final features. The final model showed high performance (recall = 98.6%, specificity = 99.7%, precision = 98.8%, accuracy = 99.6%, F1-score = 98.6%). The technology tested was shown to be efficient for monitoring cough sounds in a commercial swine production facility. It is recommended to test the technology in other environments to evaluate the effectiveness in different farm settings.

Ordinal annotation protocols for subjective visual traits implicitly assume that human evaluators can reliably anchor continuous perceptual variation to fixed discrete categories, an assumption that introduces systematic limitations in practice. Cognitive and psychophysical evidence demonstrates that this assumption is empirically unsupported. Even in controlled 'absolute identification' tasks, judgments are performed relatively, with the previous stimulus serving as the primary reference. Discrete scales therefore impose an artificial cognitive task that systematically amplifies noise, induces threshold drift, and produces inter-rater divergence that is structural rather than incidental. This paper attempts to address this limitation by replacing discrete ratings with pairwise comparisons and recovering continuous latent scales through Bradley-Terry analysis. The methodology proceeds in three stages. First, human annotators provide binary relative judgments, answering only which of two images better expresses a given trait. Second, statistical aggregation recovers a continuous latent score for each sample, achieving 86.9% consistency across approximately 56,000 image confrontations. Third, a Vision Transformer encoder is trained to map images onto a latent ordinal manifold that preserves the pairwise preference structure. Experimental results on cattle body condition scoring demonstrate that the learned geometry generalizes beyond observed comparisons, achieving pairwise classification accuracy of 96.58%, Kendall τ correlation of 0.86 with the latent reference scale, and NDCG@10 (Normalized Discounted Cumulative Gain at 10) exceeding 0.98. Structural holdout protocols (bridge- and edge-holdout) confirm that the model internalizes the trait geometry rather than memorizing local neighborhoods. These findings suggest that pairwise comparison provides a robust and principled alternative for modeling continuous perceptual traits, particularly in contexts where subjective evaluation is required.

The technologies to measure animal behavior patterns in dairy farms are important due to the possibility of continuous and accurate measuring behavior in all livestock species. There are still few validation studies using behavioral precision technologies such as behavioral monitoring collars (BMC; CowMed-Bezerra, Chip Inside Tecnologia S.A, Santa Maria, Rio Grande do Sul, BR) to monitor dairy calves. The aim of this study was to validate the BMC to assess the activity and idle behavior of dairy calves. Twenty-three Holstein dairy calves raised in a commercial farm, aged between 20 and 90 days, with an average weight of 60 ± 20 kg were randomly selected. The calves had the behaviors of rumination, feeding, idleness and activity recorded in two ways: BMC and visual observation. Visual behavior was observed at 1-minute intervals. The BMC was fitted on the calves at the 10th day after their birth. Each calf was observed for 360 min per day (from 10:00 to 13:00 h and from 19:00 to 22:00 h) over four days, totaling 1440 min. Rumination, feeding and activity times were summed and considered as general activity. General activity and idle times registered by the BMC were compared with visual observation of calves' behaviors using Pearson's correlation, correlation coefficient of agreement, linear regression, and Bland-Altman graphs for validation of precision and accuracy. Bland-Altman plots were deemed acceptable; the plots showed no bias. Bland-Altman mean differences ± standard deviation (BMC - visual observation) were -6.68 ± 2.72 min/h for general activity, and -3.86 ± 2.72 min/h for idleness. The mean differences, however, included zero within the 95% agreement interval, indicating that there was no difference between the BMC and visual observation. The slope of the linear regression did not differ from 1 for all behaviors. Therefore, idleness and general activity behaviors might be recorded precisely and accurately using BMC in dairy calves.