Veterinary medicine is undergoing a digital transformation, largely enhanced by the integration of Artificial Intelligence (AI) applications across various domains such as diagnostic imaging, disease prediction, clinical decision support, wearable monitoring, and telemedicine. This systematic survey follows the PRISMA 2020 guidelines and synthesizes research conducted from 2013 to 2025, categorizing AI's application taxonomy from convolutional neural networks to large language models, while evaluating the current landscape of AI and deep learning (DL) in veterinary digital health.
Read More
Twenty-two studies met the inclusion criteria, consisting of 18 primary evidence studies and 4 supporting domain or prototype studies. These were classified into major application domains like diagnostic imaging, predictive analytics, wearable monitoring, livestock health management, clinical decision support, and emerging LLM-based veterinary systems. The findings reveal notable advancements in areas such as automated radiography and clinical decision-making, underscoring AI's potential to improve diagnostic accuracy and clinical efficiency. However, the adoption of these technologies faces significant challenges, including fragmented datasets, species diversity, lack of extensive external validation, the complexity of many AI algorithms, and the absence of real-world clinical evaluations.
The report emphasizes the substantial promise of LLM-based applications, albeit indicating that they are still in nascent stages of implementation. It identifies key methodological limitations and barriers to adoption, calling for increased focus on standardized datasets, thorough model validation, explainable AI (XAI), and collaborative efforts among diverse stakeholders to facilitate the ethical integration of AI into veterinary practice.
Veterinary health care's digital shift is partly driven by a growing demand for telemedicine and wearable technology, illustrating the need for proactive health monitoring among pet owners. Yet, the current integration of AI remains shallow, with most published literature concentrated on algorithm development rather than clinical trials or practical implementation. Challenges such as data bias, opaque AI models, and the risks of relying excessively on automated processes contribute to continued adoption hesitancy.
The study's methodology includes a comprehensive search of peer-reviewed literature across multiple databases, targeting articles that directly apply AI tools to various facets of veterinary health. It identified a total of 169 resources, leading to 22 qualifying studies upon rigorous screening.
The study found that diagnostic imaging is the most analyzed application area, with a notable prevalence of convolutional neural networks employed for tasks such as medical image classification. The survey also highlights the lack of validated, standardized datasets in veterinary settings, a concern that limits the generalizability of findings.
As AI continues to evolve in veterinary digital health, the report highlights the need for interdisciplinary collaboration and greater emphasis on future research priorities, ensuring that AI systems are clinically relevant, safe, and effective in enhancing animal healthcare.