Researchers at the University of Cambridge have created a machine learning algorithm capable of accurately identifying heart murmurs in dogs, a significant indicator of cardiac disease, particularly prevalent among smaller breeds like King Charles Spaniels.
The research team adapted an algorithm initially designed for humans, successfully applying it to detect and classify heart murmurs in dogs using audio recordings from digital stethoscopes. In testing, the algorithm demonstrated a sensitivity of 90%, matching the accuracy of expert cardiologists.
Heart murmurs serve as a critical sign of mitral valve disease, the most common cardiac condition in adult dogs. Approximately one in 30 dogs seen by veterinarians is found to have a heart murmur, with higher prevalence among small breeds and older dogs. Early detection of mitral valve disease and other cardiac issues is essential, as timely treatment can extend the lives of affected dogs. The Cambridge-developed technology promises an affordable and efficient screening tool for primary care veterinarians, potentially improving the quality of life for countless dogs. The findings are detailed in the Journal of Veterinary Internal Medicine.
“Heart disease in humans is a significant health concern, but it poses an even greater problem in dogs,” stated Dr. Andrew McDonald, the study’s first author from Cambridge’s Department of Engineering. “Most smaller dog breeds will develop heart disease as they age. Since dogs cannot communicate their discomfort like humans, primary care veterinarians play a vital role in early detection and treatment.”
To create the algorithm, the researchers utilized a database of heart sounds from approximately 1,000 human patients. They then refined the algorithm to detect heart murmurs in dogs. Data was collected from nearly 800 dogs undergoing routine heart examinations at four veterinary specialist centers in the UK, making it the largest dataset of canine heart sounds ever compiled. All dogs received comprehensive physical examinations and echocardiograms by cardiologists to grade any heart murmurs and identify cardiac diseases.
“Mitral valve disease primarily affects smaller dogs, but to enhance our algorithm, we sought data from dogs of various sizes, shapes, and ages,” explained Professor Jose Novo Matos from Cambridge’s Department of Veterinary Medicine and co-author of the study. “The larger the dataset we can train on, the more effective our algorithm will be for veterinarians and dog owners alike.”
The team refined the algorithm to both detect and grade heart murmurs based on audio recordings, differentiating between those associated with mild disease and those indicating advanced conditions requiring treatment.
“Assessing a heart murmur and determining if treatment is needed often necessitates extensive experience and specialized tests,” said Novo Matos. “Our goal is to enable general practitioners to identify heart disease and evaluate its severity, helping pet owners make informed decisions for their dogs.”
The algorithm’s performance analysis indicated agreement with cardiologists in over half of the cases, with 90% of the assessments falling within one grade of the cardiologist’s evaluations. This is particularly promising, as variability in murmur grading among veterinarians is common.
“The grade of a heart murmur is crucial for determining subsequent treatment steps, and we have automated this process,” McDonald noted. “For veterinarians and nurses, whether they are highly skilled or less experienced with stethoscopes, this algorithm can be an invaluable tool.”
While surgery is the only treatment for humans with valve disease, effective medication options are available for dogs. “Knowing when to administer medication is critical for ensuring dogs maintain the highest possible quality of life,” said Agarwal. “Our aim is to empower veterinarians to make those decisions.”
Novo Matos added, “Many view AI as a threat to jobs, but I see it as a tool to enhance my capabilities as a cardiologist. With the current limitations on time and specialist availability, tools like this can help veterinarians and pet owners quickly identify dogs in urgent need of treatment.”
Related topics:
Bear: Australia’s Top Farm Dog Takes the Cobber Challenge Title
Guide Dog Puppies Explore Spooky Attractions Ahead of Halloween
How Dog Noses Develop Their Unique Patterns
