Manual tools for pain assessment from facial expressions have been suggested and validated for several animal species. However, facial expression analysis performed by humans is prone to subjectivity and bias, and in many cases also requires special expertise and training. This has led to an increasing body of work on automated pain recognition, which has been addressed for several species, including cats. Even for experts, cats are a notoriously challenging species for pain assessment. A previous study compared two approaches to automated ‘pain’/‘no pain’ classification from cat facial images: a deep learning approach, and an approach based on manually annotated geometric landmarks, reaching comparable accuracy results. However, the study included a very homogeneous dataset of cats and thus further research to study generalizability of pain recognition to more realistic settings is required. This study addresses the question of whether AI models can classify ‘pain’/‘no pain’ in cats in a more realistic (multi-breed, multi-sex) setting using a more heterogeneous and thus potentially ‘noisy’ dataset of 84 client-owned cats. Cats were a convenience sample presented to the Department of Small Animal Medicine and Surgery of the University of Veterinary Medicine Hannover and included individuals of different breeds, ages, sex, and with varying medical conditions/medical histories. Cats were scored by veterinary experts using the Glasgow composite measure pain scale in combination with the well-documented and comprehensive clinical history of those patients; the scoring was then used for training AI models using two different approaches. We show that in this context the landmark-based approach performs better, reaching accuracy above 77% in pain detection as opposed to only above 65% reached by the deep learning approach. Furthermore, we investigated the explainability of such machine recognition in terms of identifying facial features that are important for the machine, revealing that the region of nose and mouth seems more important for machine pain classification, while the region of ears is less important, with these findings being consistent across the models and techniques studied here.

Background Nausea and emesis can be, among other signs, common manifestations of acute vestibular system dysfunction in dogs. Currently, antiemetic drugs, such as maropitant and metoclopramide, are used commonly, but do not appear to control nausea. A non‐placebo‐controlled preliminary study suggested good efficacy of 5‐HT3‐receptor antagonists, such as ondansetron, against nausea in dogs with vestibular syndrome. Objectives To assess and confirm the effect of ondansetron on behavior suggestive of nausea in dogs with vestibular syndrome. Animals Fourteen dogs with vestibular syndrome and clinical signs of nausea presented to a neurology service. Methods Placebo‐controlled, double‐blinded, crossover study. Behavioral assessment was performed hourly for 4 hours using an established numerical rating scale. The criteria salivation, lip licking, vocalization, restlessness, lethargy, and general nausea were scored. The occurrence of emesis was recorded. After scoring at T0 (pre‐dose) and T2 (2 hours post‐dose) either ondansetron (0.5 mg/kg) or placebo was injected IV. Two hours post‐dose, treatments were switched. Blood samples were collected to measure serum arginine vasopressin (AVP) concentration, which previously has been shown to correlate with clinical signs of nausea. Results Clinical resolution of nausea was observed 1 hour after administration of ondansetron, whereas serum AVP concentration decreased 4 hours after ondansetron administration. Conclusion and Clinical Importance Administration of ondansetron IV is beneficial for dogs with nausea secondary to acute vestibular syndrome. Ondansetron substantially and rapidly decreased clinical signs of nausea behavior and stopped emesis.

Manual tools for pain assessment from facial expressions have been suggested and validated for several animal species. However, facial expression analysis performed by humans is prone to subjectivity and bias, and in many cases also requires special expertise and training. This has led to an increasing body of work on automated pain recognition, which has been addressed for several species, including cats. Even for experts, cats are a notoriously challenging species for pain assessment. A previous study compared two approaches to automated ‘pain’/‘no pain’ classification from cat facial images: a deep learning approach, and an approach based on manually annotated geometric landmarks, reaching comparable accuracy results. However, the study included a very homogeneous dataset of cats and thus further research to study generalizability of pain recognition to more realistic settings is required. This study addresses the question of whether AI models can classify ‘pain’/‘no pain’ in cats in a more realistic (multi-breed, multi-sex) setting using a more heterogenous and thus potentially ‘noisy’ dataset of 84 client-owned cats. Cats were a convenience sample presented to the Department of Small Animal Medicine and Surgery of the University of Veterinary Medicine Hannover and included individuals of different breeds, ages, sex, and with varying medical conditions/medical histories. Cats were scored by veterinary experts using the Glasgow composite measure pain scale; the scoring was then used for training AI models using two different approaches. We show that in this context the landmark-based approach performs better, reaching accuracy above 77% in pain detection as opposed to only above 65% reached by the deep learning approach. Furthermore, we investigated the explainability of such machine recognition in terms of identifying facial features that are important for the machine, revealing that the region of nose and mouth seems more important for machine pain classification, while the region of ears is less important, with these findings being consistent across the models and techniques studied here.

BackgroundIdentification of the aetiologic agent in canine discospondylitis is infrequent; and risk factors for a positive bacterial culture have not previously been reported.MethodsMedical records at three institutions were searched to identify clinical features of dogs with discospondylitis diagnosed via radiography or cross‐sectional imaging. Inclusion in this retrospective case–control study required culture of one or more samples. Multivariable binary logistic regression identified features associated with a positive culture.ResultsFifty (42%) of 120 dogs had one or more positive culture results obtained from either urine (28/115), blood (25/78), intervertebral disc aspiration (10/34) or cerebrospinal fluid (1/18). A positive culture was associated with higher bodyweight (p = 0.002, odds ratio [OR] = 1.054, 95% confidence interval [CI]: 1.019–1.089), more sample types cultured (p = 0.037, OR = 1.806, 95% CI: 1.037–3.147) and institution (p = 0.021). The presence of possibly associated preceding events (e.g., surgery), pyrexia, number of disc sites affected and serum C‐reactive protein result, among other features, were not statistically significant.LimitationsAll isolates cultured were included since differentiation of true aetiologic agents from contaminants was not possible without histological confirmation and culture from surgical or postmortem biopsies.ConclusionsClinical features typically associated with infection were not identified as risk factors for positive culture in canine discospondylitis. The statistical significance of the institution suggests that standardisation of sampling protocols is necessary.

No abstract