Clinical evaluation of the Stethee stethoscope in dogs presented for cardiac evaluation

Abstract

Approximately 10% of dogs have heart disease, most commonly myxomatous mitral valve disease (MMVD). MMVD is predisposed in certain breeds, especially the Cavalier King Charles Spaniel (CKCS). Detection of a heart murmur by cardiac auscultation is the most common technique to suspect heart disease. It is a quick, inexpensive and non-invasive method and is therefore commonly used by veterinarians.

Digital stethoscopes coupled with artificial intelligence could help the veterinarian to diagnose heart murmurs. In this thesis the use of a wireless electronic stethoscope named Stethee (M3DICINE Pty Ltd (Brisbane, Australia)) is described. It records and stores the heart sounds via Bluetooth using an App called Stethee Vet. The recordings are analysed by an Artificial Intelligence analysing system (Aida), which tries to differ recordings of animals with a normal heart sound from recordings of animals with a heart murmur.

Accuracy of Aida was assessed by comparing auscultatory findings of a veterinary cardiologist to the diagnoses made with Aida in 132 dogs. Also, the interobserver agreement between an inexperienced and experienced observer was assessed in a subset of 86 CKCS. The agreement between Aida and the experienced observer was 89% with a kappa value of 0.77 and a PPV and NPV of 83% and 93% respectively. The interobserver agreement was 88% with a kappa value of 0.70 and a PPV and NPV of 86% and 89% respectively.

The level of interobserver agreement decreased with low intensity heart murmurs, while the levels of agreement involving Aida were constant. Thus, Aida could be helpful in diagnosing low intensity murmurs for general veterinarians.