Focused Imaging as a Novel Diagnostic Strategy for Aortic Stenosis

Aortic stenosis (AS) is a valve condition that affects over 12.6 million adults and causes an estimated 102,700 deaths each year. Many patients with AS do not know about the diagnosis because it is difficult to diagnose with a stethoscope. It is estimated that there are over 560,000 undiagnosed cases of AS in the United States alone. When patients with symptomatic AS are not treated, 50% will die in 2 years. We have developed a method to automate the diagnosis of AS from cardiac ultrasound imaging using machine learning. This represents a new way to diagnose AS. In this proposal we will improve these networks to reliably identify severe AS patients that should be referred for evaluation. Additionally, we will train the networks to work with portable handheld ultrasound devices and we will study how to implement this tool in primary care offices to screen high risk patients. By developing and validating innovative machine learning (ML) methods for diagnosing AS we will establish tools to improve the identification and treatment of this life-threating condition.