Artificial Intelligence in Veterinary Medicine

Radiographs on the dyspneic cat are finally taken at midnight. The intern stares at the images, unsure of what to make of them. The pulmonary opacities are abnormal, she knows this; but their origin, and the appropriate treatment, remain a mystery. Within minutes she receives a radiology interpretation that will save the cat’s life. This interpretation did not come from an overnight teleradiology service, or an insomniac senior veterinarian, but from a computer equipped with artificial intelligence (AI).

The mention of artificial intelligence conjures images of a dystopian future worthy of a sci-fi movie: machines becoming self-aware, cyborgs battling humans for world dominance. But AI is not merely futuristic, and we should hope not dystopian. Intelligent machines are already integrated in society, from the smart phones that know our schedule to self-driving cars. And as innovation continues to evolve in every industry, researchers are also working to apply AI technology to healthcare, including veterinary medicine.

Before a discussion on the applications of artificial intelligence in veterinary medicine, it is worth noting that while the algorithms and computer systems are certainly manmade, or artificial, they are far from intelligent in the classical sense. While these systems are not self-aware or capable of novel thought, they can improve themselves automatically through experience in a process called machine learning. Their job is to make our lives easier by deducing conclusions based on input data. They are limited in the scope of the capabilities by their programming, and will likely aid, not replace healthcare professionals. Indeed, the use of AI in veterinary medicine is proving to be a boon to veterinarians and pet owners.

From finding disease to predicting it, artificial intelligence supports the value of preventive care in the clinic. Far too often, the diagnosis of a devastating disease in pets comes too late to have a meaningful impact. Now, researchers are discovering new ways to apply AI and technologically advanced learning to diagnose, treat and even predict animal heath disease through routine diagnostics performed in the clinic. As discussed in our fictitious opening scenario, artificial intelligence promises to be a disruptive force as a predictive tool for diagnostics. Systems can be trained to identify subtle deviations from normal in radiographs, which can help identify disease processes sooner. Studies on the human side have shown a high-performing algorithm may even suffice as a standalone image reader.¹ Algorithms can also aid in the identification of diseases like leptospirosis and hypoadrenocorticism through routine blood screening, increasing the value of these tests in the primary care setting.² These innovations promise earlier, more precise pet care and ultimately, better quality of life and longevity for companion animals.

The benefits of artificial intelligence in healthcare need not be restricted to the clinic. As wearable technologies gain prevalence, pet owners will benefit from data mined from collars and microchips. For example, when pet owners take advantage of Sure Petcare’s connected products such as the SureFeed® Microchip Pet Feeder Connect for cats and Animo® Activity & Behavior Tracker for dogs, they can track behavior changes and health/wellbeing indicators including activity, feeding and sleep quality, providing a detailed picture of their pet’s day-to-day routines and behaviors. These digitally connected devices can track aberrances to the routine and alert pet owners to health concerns before they become clinically apparent – and if there are changes in pets’ behavior patterns, they can review their care plan with a veterinarian. While we have known for eons that animals possess intelligence, technology and algorithms can help bridge the gap between their minds and ours, which ultimately enhances the lives of pets and those who care for them.

Still, artificial intelligence has its drawbacks and limitations. These systems are not meant to replace humans, but rather assist them. The algorithms will only be as good as their programmers, and they will need to be constantly adjusted to fit the real world. Moreover, the data supplied to these systems must be accurate and reliable to be read by the machines appropriately. The poor radiographic technique or incorrectly handled blood samples will not be interpreted appropriately by even the smartest algorithm. Intelligent veterinarians and veterinary technicians will continue to be needed to read and filter the results that an artificially intelligent, digitally connected machine produces.

For more information, please visit Merck Animal Health Intelligence or SurePetcare.com.

References

1. Hickman, S.E., Baxter, G.C. & Gilbert, F.J. Adoption of artificial intelligence in breast imaging: evaluation, ethical constraints and limitations. Br J Cancer 125, 15–22 (2021)
2. AVMA: Artificial Intelligence & Veternary Medicine (https://www.avma.org/javma-news/2020-07-15/artificial-intelligence-veterinary-medicine), Accessed August 6, 2021