Up to one-third of adult horses have valvular heart disease, which produces characteristic murmur sounds that can be detected with a stethoscope. Machine-learning algorithms combined with electronic stethoscopes have shown promise in automating the detection of murmurs and valvular heart disease in humans. However, there has been no previous work exploring the applicability of these methods to the detection of abnormal equine heart sounds.

Our approach, presented by Acoustics Lab Research Associate Andrew McDonald at the 2022 BEVA Congress, used transfer learning to apply a human-designed murmur detection algorithm to a new dataset of equine heart sounds.

More details are available on the journal page.