AI in the practice—what does the future hold?

Dr. Jonny Hughes, diagnostic imaging specialist at VetCT generating a radiology report. Radiology is one area with huge potential for AI to improve efficiency and augment the radiologist. Courtesy Dr. Liz Barton Artificial intelligence (AI) is increasingly impacting both clinical and non-clinical areas of veterinary practice. With its rapid pace of development, a void of uncertainty about the future impact on veterinary teams and the care they provide is becoming apparent. However, by extrapolating from human medicine and wider industries, we can project the path and, ideally, shape the development of AI to benefit veterinary teams and animal health, while proactively avoiding some of the potential pitfalls. AI—the basics There are a few fundamentals of AI that help to frame the discussion. First, AI refers to the simulation of human intelligence in machines designed to perform tasks that typically require human intelligence, such as visual perception, speech recognition, decision-making, and language translation. Fundamentally, it involves feeding data into computer systems to enable problem solving. Rather than one field, AI comprises a series of subfields including machine learning and deep learning, through which algorithms make predictions or classifications based on input data. This is a key concept—the outputs can only be as good as the data they receive, and in the veterinary field, access to high quality, standardized, representative data can present a significant challenge. The launch of OpenAI’s ChatGPT to the masses in November 2022 triggered significant and ongoing media hype about all things AI. However, it may be surprising to learn the technology is still limited to the field of “weak” AI—also called narrow AI or artificial narrow intelligence (ANI)—as defined by its design to perform specific tasks. The term “weak” belies the power of this technology to enable robust applications, such as Apple’s Siri, Google’s Bard, image-generator Midjourney, and self-driving vehicles. “Strong” AI refers to artificial general intelligence (AGI)—also known as general intelligence—where machines would have intelligence equal to humans and possess a self-aware consciousness with the ability to solve problems, learn, and plan for the future. Beyond this lies the science-fiction field of artificial super intelligence (ASI), or superintelligence, where it would surpass the intelligence and ability of the human brain – the stuff of Hollywood movies. For now, strong AI remains entirely theoretical, but seeing the exponential growth in the learning capacity of ChatGPT alone, it is not outside the bounds of possibility in the near future (estimates for its arrival have been cut in more than half; from 50 years to 20 years). The arrival of general AI represents a singularity point beyond which, whether it is truly conscious or not, a machine could continuously improve itself and reach far beyond human capabilities. Proactive regulation to control this point is the impetus driving many countries to act rapidly and cooperatively to ensure they retain control, rather than the AI developers. AI in your practice Today, the breadth of narrow AI is expanding vastly through a variety of applications that could impact veterinary practice. Even before veterinary patients arrive at the clinic, patient triage and appointment booking could be facilitated through smart scheduling systems. Such systems could learn to anticipate and factor individual preferences into scheduling, such as client and clinician communications styles, skill sets and task preferences. For example, the veterinarian who likes dentals could be automatically allocated more dental procedures, and the veterinary technician with imaging expertise could be scheduled to work in radiology on the busiest diagnostic imaging days. These systems could also adapt quickly to change; for example, if a veterinarian calls in sick, the system could automate rescheduling appointments and dispatch communications to both clients and team members. On arrival, traditional clinic reception may be replaced with more of a concierge service to meet and greet, putting patients and clients at ease. Rather than tending to calls, payments, and bookings, a concierge would have the time to build relationships, answer questions, and promote additional products and services, while a machine handles the other tasks. During—or even prior to—the veterinary consultation, data from wearable devices and remote monitors could be presented to the veterinary team, highlighting trends or deviations alongside recommendations, adding context to the clinical examination. For the business, improved data gathering and analysis could provide an array of insights including pharmaceutical use, financials and performance, equipment use, and the physical flow of people and animals within clinic areas. Following the trends in smart technology-enabled homes, smart practices could optimize use of energy, reduce waste and conserve water, improving sustainability, and reducing costs. Hospitals could even use AI to develop digital “twins” of themselves to simulate and test changes like these to their environment, ensuring efficacy before implementing them. The virtual practice could also be used to provide owners with the option to live monitor their pet’s progress through the virtual environment. Deployed intelligently, AI in the practice could be used to create smarter, more connected, efficient, and profitable clinics, while providing a more personalized service to clients. AI and your team The great hope for AI is it helps solve some of the workforce challenges of stress, burnout, recruitment, and retention by improving working life. Administrative burden is recognized as contributing to burnout in medical professionals, and this is one area AI is primed to help.1 Improving the well-being of veterinary team members is an important focus and should be a central consideration with any innovation. Automated transcription of clinical discussions into clinical notes that integrate with electronic health records enable the clinician to focus more on the patient and owner, without having to divert time and effort to accurately document the interaction. Tools can also learn the style and preferences of individual clinicians and integrate expert subject matter for further context.2 Already used in human healthcare, this could also be deployed in veterinary clinics today. Chatbots enable smart owner communications, again adapting to the style and preference of the clinician and the owner. Recently launched tools pull relevant data from the clinical record to formulate chat responses to owners while ensuring there is a human in the loop; it is free to edit and adapt the text.3 Identifying inefficiencies and improving workflows, such as digitization and automated completion of paperwork, should free up veterinarians’ and technicians’ time to do the skilled, motivating, clinical tasks they are trained for, representing a better use of human resources. Clinical confidence can be bolstered with a virtual second opinion, though the tendency to believe machine over human—known as automation bias—must be considered where opinion differs.4 On-the-job training tools (such as virtual and augmented reality programs) can promote patient-side skills development for both clinical procedures and non-clinical skills, such as communication to both clients and colleagues.5 While there may be concern AI could replace veterinary team members, it is extremely unlikely the nuanced interactions within the veterinarian-client-pet relationship will ever be fully achieved by an AI tool. However, practices that embrace the technology, improving working life, clinical outcomes, and cost and time efficiencies, will certainly have competitive advantages over those that do not. Patients From genetic selection to informed breeding to end-of-life care, AI has huge potential to improve health and welfare. Indeed, advancing healthcare is perhaps the most important area to quickly, effectively, and responsibly utilize non-human intelligence. With human medicine far in advance of the veterinary field with this technology, the application of AI in One Health has enormous potential to benefit both humans and animals. For example, image analysis of human facial expressions have been mapped on to other species, including horses.6 Paired with remote, continuous monitoring, this could pave the way for early detection of facial expressions of pain in hospitalized equines without the risk of behavior adaptation in the presence of a human observer. Automated lameness detection in dairy cattle through gait analysis is already enabling early intervention, especially in large herds where observation time per animal is limited.7 Scaled up, the potential welfare gains to many species for lameness and pain management alone are significant. While we are some way from precision medicine even in human healthcare, the potential to individualize care pathways throughout the lifespan of a pet is exciting. Continuous monitoring in the home through wearable sensors and remote observation may reduce the need for travel and hospitalization. When ECG, respiratory rate and pattern, activity levels, food and water intake, weight, blood parameters, etc. can be monitored remotely, there is less need for invasive or in-clinic monitoring with patients potentially being discharged from hospitals earlier for at-home care. Veterinary patients can also benefit from the huge strides in technology being developed for human medicine. The time (and cost) taken to decode genetic information is a fraction of what it was just 20 years ago when the human genome project was completed—now taking hours rather than months. Comparative genomics juxtapose genes from different species to analyze for disease associations. Some of the most powerful applications of AI to drug discovery were accidental outputs from AI, such as the ability to predict protein folding patterns from genetic code, rapidly accelerating the selection process for new potential therapeutics.8 The opportunities are truly beyond conception. One estimation of the evolution of AI applications in human medicine show rapid progress within a decade:9 0-5 years—Automation of time consuming, high-volume repetitive tasks 5-10 years—Use of multi-modal datasets to drive precision therapeutics >10 years—Precision medicine through AI-augmented healthcare and connected care Clients AI offers owners huge potential gains in knowledge and understanding of a pet’s health and welfare. The modern pet parent is increasingly integrating their pet fully into family life. As the wealth of health data that is available to humans through smartphones and wearables is becoming more incorporated to inform their own daily life and choices, this trend for self-informed healthcare is bound to accelerate rapidly for their pets. Even before owning a pet, AI could be used to help select the right pet for the lifestyle and circumstances of a potential owner, matching them to a species and breed, linking to breeders and providing advice about how to care for their new pet. Once in the home, the mutual relationship between animal and owner can be greatly enhanced through AI tools. The non-verbal signals animals produce can be decoded through AI, with pattern recognition of visual and auditory data being translated into outputs we can understand and act upon. For example, auditory data is used to provide early warning signs of stress in chickens from subtle changes in their vocalizations.10 Continuous monitoring through wearable sensors and image capture can provide the owners with understanding of clinical conditions, nutrition, activity, and behavior. Automated notifications can provide early warning of changes, such as changes in activity due to pain or increased water intake in a diabetic. However, there are significant risks associated with clinical tools accessible to pet owners. For example, an image analysis screening for ocular disease in the anterior chamber of horses may be useful to augment exam by a clinician, who can also assess the posterior chamber, peri-ocular environment and the animal as a whole. However, an owner using such a tool to check for “eye disease” could misinterpret an “all clear” for this one region of the eye to mean ocular disease is not present, potentially delaying urgent treatment and leading to negative outcomes.11 Veterinarians and AI developers need to work closely together to ensure the responsible deployment and use of clinical tools in the public domain. A key benefit of AI will be to increase access to care for more pets and owners. Any owner with an internet-connected device can potentially interact with AI-driven information and education tools that lower the point of entry to veterinary care. However, it is vitally important to ensure there is a veterinarian in the loop to guide the appropriate application of the technology. Conclusion This is a moment in time when the technology for capturing, storing and assessing multi-modal data is accelerating ever more rapidly. AI can only exist with data to inform its development, and while this may provide a barrier to the development of veterinary tools today, we will soon see AI being used to extract and standardize data sources to fuel a second, more rapid wave of innovation in the industry. Learning the boundaries of how to apply AI in practice now, as well as how to educate and guide clients and ensuring veterinary team members are empowered with the knowledge and skills to deploy AI effectively, will be crucial to foster trust, confidence, enthusiasm and responsible use of this technology. Certain job roles will need to change and adapt as AI tools become more widespread, but it is extremely unlikely the human-animal and human-human bonds and the empathy at the core of veterinary practice will ever be replaced. However, change management to ease the adoption of the technology across teams will be important. To make AI, humans must clearly identify and define the problem that needs solving, rather than creating clever AI tools and retrofitting them to different problems. To do this, we must engage with stakeholders and advocate for those without a voice (our patients!), considering needs, workflow integrations and ethical implications (such as the impact on jobs). We then need to source the unbiased and representative data to feed the algorithm during the learning process. Once solutions are created, these need to be tested, validated, and refined, with follow-up surveillance of performance and safety. In addition, it is vital veterinarians are influencing the development of technologies that impact animal health and welfare to ensure they provide a benefit. The profession must also work on creating robust guidelines and standards to apply to AI tools to ensure they are accurate and reliable. Veterinary care will inevitably shift towards a more preventative, individualized approach, with practice life centered more around the relational aspect of pet and client interactions and clinical skills. This is often the most satisfying aspect of clinical practice, and with the potential for AI-informed practice to improve patient outcomes, the future of AI-enabled veterinary practice is bright. Liz Barton, MA, VetMB, MRCVS, MCIPR, is a veterinarian and head of communications at VetCT, a company providing veterinary teleradiology, teleconsulting and education services to veterinary clinics and universities worldwide. She is passionate about exploring topics that impact the wellbeing of veterinary teams and animal health and welfare. Dr. Barton recently co-authored the white paper, Artificial Intelligence in Veterinary Medicine, providing a foundational understanding of AI and exploring how it can impact the veterinary ecosystem. References Chaiyachati KH, Shea JA, Asch DA, Liu M, Bellini LM, Dine CJ, et al. 2019. Assessment of inpatient time allocation among first-year internal medicine residents using time-motion observations. JAMA Int Med. 179:760–7. doi: 10.1001/jamainternmed.2019.0095 https://www.nuance.com/healthcare/ambient-clinical-intelligence.html https://petsapp.com/blog/petsapp-copilot-the-first-ai-communication-assistant-for-veterinary-teams Vered M et al. 2023. The effects of explanations on automation bias, Artificial Intelligence, Volume 322, 103952, ISSN 0004-3702, https://doi.org/10.1016/j.artint.2023.103952. https://nurtur.co/ Andersen PH, Broomé S, Rashid M, Lundblad J, Ask K, Li Z, Hernlund E, Rhodin M, Kjellström H. 2021. Towards Machine Recognition of Facial Expressions of Pain in Horses. Animals (Basel). Jun 1;11(6):1643. doi: 10.3390/ani11061643. PMID: 34206077; PMCID: PMC8229776 Viazzi S., Bahr C., Schlageter-Tello A., Van Hertem T., Romanini C.E.B., Pluk A., Halachmi I., Lokhorst C. & Berckmans D. (2013). – Analysis of individual classification of lameness using automatic measurement of back posture in dairy cattle. J. Dairy Sci., 96 (1), 257–266 https://www.nature.com/articles/d41586-020-03348-4 Bajwa J, Munir U, Nori A, Williams B. Artificial intelligence in healthcare: transforming the practice of medicine. Future Healthc J. 2021 Jul;8(2):e188-e194. doi: 10.7861/fhj.2021-0095. PMID: 34286183; PMCID: PMC8285156. Pourdarbani, R., Yarahmadi, P., & Maleki, S. (2021). FowlGuard: An IoT-based smart system for early detection of poultry diseases. Computers and Electronics in Agriculture, 182, 105983 May, A, Gesell-May, S, Müller, T, Ertel, W. 2021. Artificial intelligence as a tool to aid in the differentiation of equine ophthalmic diseases with an emphasis on equine uveitis. Equine Vet J. 54: 847–855. doi:10.1111/evj.13528

Cookie	Duration	Description
cookielawinfo-checkbox-analytics	11 months	This cookie is set by GDPR Cookie Consent plugin. The cookie is used to store the user consent for the cookies in the category "Analytics".
cookielawinfo-checkbox-functional	11 months	The cookie is set by GDPR cookie consent to record the user consent for the cookies in the category "Functional".
cookielawinfo-checkbox-necessary	11 months	This cookie is set by GDPR Cookie Consent plugin. The cookies is used to store the user consent for the cookies in the category "Necessary".
cookielawinfo-checkbox-others	11 months	This cookie is set by GDPR Cookie Consent plugin. The cookie is used to store the user consent for the cookies in the category "Other.
cookielawinfo-checkbox-performance	11 months	This cookie is set by GDPR Cookie Consent plugin. The cookie is used to store the user consent for the cookies in the category "Performance".
JSESSIONID	session	The JSESSIONID cookie is used by New Relic to store a session identifier so that New Relic can monitor session counts for an application.
viewed_cookie_policy	11 months	The cookie is set by the GDPR Cookie Consent plugin and is used to store whether or not user has consented to the use of cookies. It does not store any personal data.

Cookie	Duration	Description
vuid	2 years	Vimeo installs this cookie to collect tracking information by setting a unique ID to embed videos to the website.
_ga	2 years	The _ga cookie, installed by Google Analytics, calculates visitor, session and campaign data and also keeps track of site usage for the site's analytics report. The cookie stores information anonymously and assigns a randomly generated number to recognize unique visitors.
_gat	1 minute	This cookie is installed by Google Universal Analytics to restrain request rate and thus limit the collection of data on high traffic sites.
_gat_UA-	1 minute	A variation of the _gat cookie set by Google Analytics and Google Tag Manager to allow website owners to track visitor behaviour and measure site performance. The pattern element in the name contains the unique identity number of the account or website it relates to.
_gid	1 day	Installed by Google Analytics, _gid cookie stores information on how visitors use a website, while also creating an analytics report of the website's performance. Some of the data that are collected include the number of visitors, their source, and the pages they visit anonymously.
__gads	1 year 24 days	The __gads cookie, set by Google, is stored under DoubleClick domain and tracks the number of times users see an advert, measures the success of the campaign and calculates its revenue. This cookie can only be read from the domain they are set on and will not track any data while browsing through other sites.

Cookie	Duration	Description
ADV_u_id	3 months 8 days	Unique customer identifier used to track unique ad views and interactions with some ads.
anj	3 months	AppNexus sets the anj cookie that contains data stating whether a cookie ID is synced with partners.
fr	3 months	Facebook sets this cookie to show relevant advertisements to users by tracking user behaviour across the web, on sites that have Facebook pixel or Facebook social plugin.
IDE	1 year 24 days	Google DoubleClick IDE cookies are used to store information about how the user uses the website to present them with relevant ads and according to the user profile.
loc	never	AddThis sets this geolocation cookie to help understand the location of users who share the information.
OAGEO	session	OpenX sets this cookie to avoid the repeated display of the same ad.
OAID	1 year	Cookie set to record whether the user has opted out of the collection of information by the AdsWizz Service Cookies.
test_cookie	15 minutes	The test_cookie is set by doubleclick.net and is used to determine if the user's browser supports cookies.
uuid2	3 months	The uuid2 cookie is set by AppNexus and records information that helps in differentiating between devices and browsers. This information is used to pick out ads delivered by the platform and assess the ad performance and its attribute payment.
_fbp	3 months	This cookie is set by Facebook to display advertisements when either on Facebook or on a digital platform powered by Facebook advertising, after visiting the website.

AI in the practice—what does the future hold?

The impact of AI on the veterinary industry and guidelines for responsible development and use

Membership Required