As reported in the Journal of Veterinary Internal Medicine, researchers from the University of Cambridge have developed a machine learning algorithm to detect heart murmurs in dogs, a key indicator of cardiac disease that significantly affects certain small breeds, such as King Charles Spaniels.
Image Credit: Tatyana Vyc/Shutterstock.com
The research team has adapted an algorithm initially designed for humans, demonstrating that it could accurately detect and grade heart murmurs in dogs using audio recordings from digital stethoscopes. In tests, the algorithm achieved a 90 % sensitivity rate, matching the accuracy of expert cardiologists.
Heart murmurs are a primary indicator of mitral valve disease, the most prevalent heart condition in adult dogs. Approximately one in 30 dogs examined by veterinarians present with a heart murmur, though the condition is more common in smaller and older breeds.
Given the high prevalence of mitral valve disease and other cardiac conditions in dogs, early detection is essential, as timely intervention can extend their lives. The Cambridge-developed technology could become an affordable and effective screening tool for primary care veterinarians, ultimately enhancing dogs' quality of life.
Heart disease in humans is a huge health issue, but in dogs it’s an even bigger problem. Most smaller dog breeds will have heart disease when they get older, but obviously dogs can’t communicate in the same way that humans can, so it’s up to primary care vets to detect heart disease early enough so it can be treated.
Dr. Andrew McDonald, Study First Author and Research Associate, Department of Engineering, University of Cambridge
Professor Anurag Agarwal, who led the research, is a specialist in acoustics and bioengineering.
As far as we are aware, there are no existing databases of heart sounds in dogs, which is why we started out with a database of heart sounds in humans. Mammalian hearts are fairly similar, and when things go wrong, they tend to go wrong in similar ways.
Anurag Agarwal, Professor, University of Cambridge
The researchers began with a heart sound database from approximately 1000 human patients, developing a machine-learning algorithm to replicate cardiologist-detected heart murmurs. They subsequently adapted this algorithm for use with canine heart sounds.
To validate the adapted algorithm, the team collected data from nearly 800 dogs undergoing routine heart exams at four veterinary specialist centers in the UK.
Each dog received a comprehensive physical exam and an echocardiogram by a cardiologist to assess heart murmurs and diagnose cardiac disease, with heart sounds recorded using an electronic stethoscope. This dataset is, by an order of magnitude, the largest ever created for canine heart sounds.
Mitral valve disease mainly affects smaller dogs, but to test and improve our algorithm, we wanted to get data from dogs of all shapes, sizes and ages. The more data we have to train it, the more useful our algorithm will be, both for vets and for dog owners.
Jose Novo Matos, Study Co-Author and Professor, Department of Veterinary Medicine, University of Cambridge
The researchers fine-tuned the algorithm to not only detect but also grade heart murmurs from audio recordings, enabling it to distinguish between murmurs associated with mild disease and those indicating advanced heart disease requiring further treatment.
Novo Matos added, “Grading a heart murmur and determining whether the heart disease needs treatment requires a lot of experience, referral to a veterinary cardiologist, and expensive specialized heart scans. We want to empower general practitioners to detect heart disease and assess its severity to help owners make the best decisions for their dogs.”
Performance analysis revealed that the algorithm aligned with the cardiologist’s assessment in over half of cases and was within one grade of the cardiologist’s evaluation in 90 % of cases. The researchers view this as a promising result, particularly given the usual variability in heart murmur grading among veterinarians.
Dr. McDonald stated, “The grade of heart murmur is a useful differentiator for determining next steps and treatments, and we have automated that process. For vets and nurses without as much stethoscope skill, and even those who are incredibly skilled with a stethoscope, we believe this algorithm could be a highly valuable tool.”
In humans with valve disease, the only treatment is surgery, but for dogs, effective medication is available.
Agarwal noted, “Knowing when to medicate is so important, in order to give dogs the best quality of life possible for as long as possible. We want to empower vets to help make those decisions.”
“So many people talk about AI as a threat to jobs, but for me, I see it as a tool that will make me a better cardiologist. We can’t perform heart scans on every dog in this country – we just don’t have enough time or specialists to screen every dog with a murmur. But tools like these could help vets and owners, so we can quickly identify those dogs who are most in need of treatment,” Novo Matos stated.
The Medical Research Council, Emmanuel College Cambridge, and the Kennel Club Charitable Trust provided partial funding for the study.
Journal Reference:
McDonald, A., et. al. (2024) A machine-learning algorithm to grade heart murmurs and stage preclinical myxomatous mitral valve disease in dogs. Journal of Veterinary Internal Medicine. doi.org/10.1111/jvim.17224