New research by the Acoustics Group has been published in the Journal of Veterinary Internal Medicine (JVIM).
Heart murmurs are useful diagnostic indicators of heart disease in dogs, particularly small breeds that are predisposed to myxomatous mitral valve disease (MMVD). However, accurately detecting and grading heart murmurs requires significant skill, and previous studies have demonstrated substantial inter-observer variability.
In this research, we collaborated with four veterinary referral centres in the UK to collect a world-leading dataset of heart sounds from 756 dogs. We then trained a machine learning algorithm to grade the presence of heart murmurs in the electronic stethoscope recordings. On withheld testing, the algorithm was able to both accurately grade murmurs and predict the stage of asymptomatic mitral valve disease.
This research holds promise for improving early detection and management of mitral valve disease in dogs, potentially reducing unnecessary referrals for dogs with only mild conditions.
The research was featured on BBC News and by the central University of Cambridge.
Andrew McDonald et al. ‘A machine learning algorithm to grade canine heart murmurs and stage preclinical myxomatous mitral valve disease.’ Journal of Veterinary Internal Medicine (2024). DOI: 10.1111/jvim.17224
Photo credit: Jacqueline Garget
