Position Statements

Artificial Intelligence in Veterinary Medicine

October 12, 2023

Artificial Intelligence and its application to veterinary medicine is a highly dynamic area that will continue to evolve beyond the publication date of this position. Currently (2023) there are a number of purported benefits provided by tools and services based on artificial intelligence, however readers should be aware that such claims may not always be supported by sound evidence. The Canadian Veterinary Medical Association (CVMA) advises readers to contact their respective provincial or territorial regulatory bodies for the latest on relevant standards and policies for the use of AI technologies within their jurisdictions.

Background

1. Artificial intelligence (AI) is a branch of computer science in which computer systems are designed to perform tasks that mimic human intelligence. AI is a broad umbrella term that encompasses a variety of subfields and techniques including machine learning, deep learning, natural language processing, pathomics, and radiomics (1,2).
2. Generative AI (e.g. large language models) is increasingly playing a role in many medical and non-medical applications. The rapid development and improvement of generative AI models demonstrates the potential role of AI to develop training data for broad application in veterinary medicine as well as create generated images, including applications with computer vision and radiomics (3).
3. In human medicine, AI is being used for medical image interpretation in radiology, pathology, ophthalmology, and gastroenterology as well as advanced applications in clinical object detection during surgery and the development of pathology biomarkers for disease detection (4).
4. Currently (2023) a wide range of AI applications are being employed in veterinary medicine. Some of the ways AI is currently being used by veterinary professionals with the aim of benefiting their clients and patients include:
   
   • Analysis of large datasets (health records including patient records, laboratory data, medical images, pathologic specimens, etc.) with the aid of computer algorithms to help make or improve diagnoses, prognosis prediction, therapy, and patient outcomes, and streamline administrative tasks (2,4).
   • Diagnostic imaging: to quickly analyze medical images, such as radiographs and ultrasound scans, to help diagnose conditions and guide treatment decisions (5).
   • Pathology/ hematology/ parasitology: to provide automated blood cell differentials, urine sediment examination and fecal analysis with the aim of increased efficiency and accuracy.
   • Clinical decision support: to provide real-time support to practitioners in the form of early diagnosis, diagnostic algorithms, differential diagnoses, treatment recommendations, and drug dosages (6).
   • Disease detection and surveillance: to quickly detect and track outbreaks of disease in animal populations, allowing for early intervention and control (7, 8, 9).
   • Automated livestock monitoring: to monitor livestock health and behavior, providing early warning signs of illness or distress and allowing for early intervention. Analysis of animal production data (weight gain/loss, feed consumption, mild production volumes etc.) and analysis of electronic health records. (10, 11).
   • Predictive modeling: to assist with drug development (4, 11) (e.g. simulation platforms) and to model the spread of diseases in animal populations, allowing for better understanding of disease transmission and more effective control strategies (10, 11).
   • Antimicrobial resistance: to help identify and track the emergence and spread of antimicrobial-resistant bacteria (12).
5. Despite well-described purported benefits of AI, risks associated with the use of AI in veterinary medicine should be carefully considered, including:
   • Bias in data sets, and selection of algorithms: AI algorithms can reflect the biases in data sets they are trained on, leading to incorrect diagnoses or treatment recommendations.
   • Lack of transparency: The decision-making processes of AI systems can be opaque, making it difficult to understand how they arrived at their conclusions. 
   • Inaccurate results: AI systems in most cases have not been formally and independently evaluated. Since they are only as accurate as the data they are trained on, errors in training data can lead to inaccurate interpretation in a clinical context.
   • Dependence on technology: Relying on AI for diagnoses and treatments can reduce critical thinking and problem-solving skills in veterinary professionals.
   • Ethical concerns: AI has the potential to create ethical dilemmas by replacing human decision-making with automated processes (13).
6. AI for use in human medicine is currently (2023) considered Software as a Medical Device and is subject to approval by the FDA in the United States and Health Canada in Canada. As AI in human medicine is predominantly the domain of specialists alone, professional bodies such as the Canadian Association of Radiologists have issued white papers (14) and position statements on AI. This oversight provides guidance to the medical profession and ensures the safety of patients prior to rapid adoption or commercialization.
7. Regarding veterinary diagnostics in Canada, there is currently a lack of regulatory oversight, meaning that commercial AI systems for veterinary diagnostics may be brought to market with little or no supporting evidence (15).
8. Though Canada developed a national strategy on AI in 2017, a comprehensive set of international AI standards and conformity assessment has yet to fully emerge (15).
9. As a current example, computer-aided diagnostic systems are being employed by veterinary hospitals across Canada to interpret radiographs. While these systems are commercially available, there is little to no open-access information on the underlying data sources and algorithms available to veterinary professionals.
10. Veterinary professionals should be aware that lack of structured regulatory oversight could result in liability for veterinary practitioners in the event that errors occur as a result of reliance on AI systems.
11. Veterinary professionals should be aware that veterinary platforms based on human systems may not be translatable to animals.
12. The CVMA holds that AI has the potential to enhance the quality of care for animals, improve efficiency in the delivery of veterinary services, and provide new insights into the underlying mechanisms of animal diseases. However, it is important to ensure that AI is used in a responsible and ethical manner, and that its benefits are balanced with considerations for bias and misdiagnosis risks, animal welfare, and privacy (14).
13. The CVMA encourages regulatory bodies to develop policies and standards for the use of AI applications used in veterinary medicine to mitigate risk to veterinary professionals, their clients and patients, and to promote the use of good machine learning practices (16,17).
14. Veterinary professionals should refer to existing and new provincial or territorial policies concerning AI systems and their applications in practice. A list of provincial regulatory bodies can be found on the CVMA website (18).

References

1. American Veterinary Medical Association. 2020. Artificial intelligence & veterinary medicine. https://www.avma.org/javma-news/2020-07-15/artificial-intelligence-veterinary- medicine
2. Appleby RB, Basran PS. Artificial intelligence in veterinary medicine. J Am Vet Med Assoc. 2022 Mar 30;260(8):819-824. https://avmajournals.avma.org/view/journals/javma/260/8/javma.22.03.0093.xml
3. Sorin V, Barash Y, Konen E, Klang E. Creating Artificial Images for Radiology Applications Using Generative Adversarial Networks (GANs) - A Systematic Review. Acad Radiol. 2020 Aug;27(8):1175-1185. DOI: 1016/j.acra.2019.12.024
4. Rajpurkar, P., Chen, E., Banerjee, O. et al. AI in health and medicine. Nat Med 28, 31–38 (2022). https://doi.org/10.1038/s41591-021-01614-0
5. Hennessey E, DiFazio M, Hennessey R, Cassel N. Artificial intelligence in veterinary diagnostic imaging: A literature review. Vet Radiol Ultrasound. 2022 Dec;63 Suppl 1:851-870. DOI: 1111/vru.13163
6. Giordano Chris, Accessing Artificial Intelligence for Clinical Decision-Making. Frontiers in Digital Health. 3. 2021 https://www.frontiersin.org/articles/10.3389/fdgth.2021.645232. DOI=10.3389/fdgth.2021.645232
7. Ezanno, P., Picault, S., Beaunée, G. et al. Research perspectives on animal health in the era of artificial intelligence. Vet Res 52, 40 (2021). https://doi.org/10.1186/s13567-021-00902-4
8. Guitian J, Arnold M, Chang Y, Snary EL. Applications of machine learning in animal and veterinary public health surveillance. Rev Sci Tech. 2023 May;42:230-241. English. DOI: 20506/rst.42.3366
9. Reagan KL, Deng S, Sheng J, et al. Use of machine-learning algorithms to aid in the early detection of leptospirosis in dogs. Journal of Veterinary Diagnostic Investigation. 2022;34(4):612-621. doi: 1177/10406387221096781
10. Jun Bao, Qiuju Xie Artificial intelligence in animal farming: A systematic literature review, Journal of Cleaner Production, Volume 331, 2022, 129956, ISSN 0959-6526, https://doi.org/10.1016/j.jclepro.2021.129956.
11. Andreas Bender, Isidro Cortés-Ciriano, Artificial intelligence in drug discovery: what is realistic, what are illusions? Part 1: Ways to make an impact, and why we are not there yet, Drug Discovery Today, Volume 26, Issue 2, 2021,Pages 511-524,ISSN 1359-6446, https://doi.org/10.1016/j.drudis.2020.12.009
12. Rabaan AA, et al. Application of Artificial Intelligence in Combating High Antimicrobial Resistance Rates. Antibiotics (Basel). 2022 Jun 8;11(6):784. doi: 3390/antibiotics11060784
13. Cohen EB, Gordon EC First, do no harm. Ethical and legal issues of artificial intelligence and machine learning in veterinary radiology and radiation oncology, Veterinary Radiology and Ultrasound. First published: 13 December 2022 https://doi.org/10.1111/vru.13171
14. Tang A, Tam R, Cadrin-Chênevert A, et al. Canadian Association of Radiologists White Paper on Artificial Intelligence in Radiology. Canadian Association of Radiologists Journal. 2018;69(2):120-135. doi:1016/j.carj.2018.02.002
15. Standards Council of Canada . 2023 . Discerning signal from noise: The state of global AI standardization and what it means for Canada. https://www.scc.ca/en/system/files/publications/SRI_DiscerningSignalFromNoise_English_v2.pdf 
16. American College of Veterinary Radiology. Artificial Intelligence https://acvr.org/artificial-intelligence-in-veterinary-diagnostic-imaging-and-radiation-oncology/
17. USFDA, Health Canada 2021. Good Machine Learning Practice for Medical Device Development: Guiding Principles https://www.fda.gov/media/153486/download
18. CVMA, Regulatory Bodies https://www.canadianveterinarians.net/public-resources/regulatory-bodies/