The Accuracy of an Artificially-intelligent Stethoscope

Congenital Heart Disease, Heart Murmurs, Structural Heart Disease

Murmur, Artificial-Intelligence, Stethoscope, Congenital Heart Disease, Structural Heart Disease

This study will characterize the accuracy of a commercially available artificially-intelligent stethoscope in determining which childhood murmurs suggest underlying congenital structural heart disease and therefore warrant diagnostic echocardiograms.

Heart murmurs can be ausculted in most pediatric patients, but underlying Congenital Structural Heart (CSH) Disease is rare. Nevertheless, CSH Disease is often first suggested by the presence of a murmur. Primary care providers refer pediatric patients to Cardiologists for murmur evaluations with low accuracy, and this pattern results in: 1. needless emotional strain on many patients with functionally normal hearts and their families, 2. increased morbidity and mortality rates for rare patients with delayed cardiac diagnoses, and 3. exorbitant costs to society. Referral accuracy for murmur evaluations probably is low because the differences between innocent and pathological murmurs can be subtle. Computer-assisted analysis of heart sounds may increase the accuracy of primary care referrals to pediatric cardiologists, but the accuracy of the only FDA-approved artificially-intelligent cardiac auscultation system, Zargis Medical Corporation's Cardioscan®, has never been determined prospectively in an unselected population of live pediatric patients. As a primary endpoint, the study herein proposed would prospectively compare the sensitivity and specificity of the Cardioscan® with that of CHMCA's Pediatric Cardiologists in the identification of cardiac pathology among 300 new and unselected pediatric patients referred to CHMCA's cardiologists for murmur evaluations, using two-dimensional transthoracic echocardiography as the common gold-standard. Double-blinding will be in place. The study also will attempt secondarily to gauge the clinical significance of pathology missed by the Cardioscan®, to define sub-populations of patients among which the Cardioscan's® judgment should be considered less accurate, and to gather data concerning the value and acceptability to families of the Cardioscan®. If (as expected) the Cardioscan's® accuracy merely approaches that of Pediatric Cardiologists, then the study herein proposed would still establish the Cardioscan® as a formidable tool for primary care providers-a helpful, billable, and potentially cost-saving alternative to consulting Cardiology on patients with probably- or possibly-innocent murmurs. Data collection for this project will occur predominantly within CHMCA's Heart Center in Akron, although some data collection may occur within CHMCA and/or at CHMCA's Heart Centers in Beachwood and/or Boardman. Minimal disruption to existing clinical patient flow through those settings would be expected and the entire protocol is expected to be completely painless and noninvasive, with no foreseeable health risks.

No related papers found yet.

United States