Sign in Subscribe
Beth Israel Deaconess Medical Center

New Artificial Intelligence Tool Successfully Detects Severe Aortic Stenosis

Strategic Insights

— 1 min read

Researchers at Beth Israel Deaconess Medical Center (BIDMC) have successfully validated an artificial intelligence tool that detects severe aortic stenosis (AS) with over 98 percent accuracy. The tool, artificial intelligence decision support algorithm (AI-DSA), uses routine echocardiographic reports to predict an individual's probability of having severe AS. This breakthrough has the potential to improve the identification of individuals with severe AS symptoms and subsequent referral for lifesaving aortic valve replacement (AVR).

Key Takeaways:

  • The AI-DSA tool was tested using routine echocardiographic reports from over 31,000 U.S. Medicare beneficiaries at BIDMC, demonstrating an accuracy rate of over 98% in detecting severe AS.
  • The tool identified 98% of patients who met clinical guidelines for severe AS and identified over 1,000 patients who likely had severe AS, but did not meet clinical guidelines.
  • Patients who likely had severe AS, but did not meet clinical guidelines, had the same risk of death as those with severe AS, yet only 6.6% received AVR, compared to 20.2% of people diagnosed with severe AS.
  • The study suggests the AI-DSA tool can be used alongside standard echocardiographic reporting to improve the detection of severe AS and subsequent referral for AVR.
  • The AI-DSA tool was originally trained and tested using over 1 million echocardiograms from the National Echocardiographic Database of Australia.
  • The study highlights the potential for the AI-DSA tool to improve clinical practice and overall detection of individuals with severe AS, with the tool undergoing clinical testing to ensure its effectiveness.

Statistics:

  • Over 98% accuracy rate in detecting severe aortic stenosis (AS) using AI-DSA tool.
  • 1,000 patients who likely had severe AS, but did not meet clinical guidelines.
  • 6.6% of patients who likely had severe AS, but did not meet clinical guidelines, received AVR.
  • 20.2% of patients diagnosed with severe AS received AVR.
  • Over 1 million echocardiograms from the National Echocardiographic Database of Australia used to train and test AI-DSA tool.
  • 31,000 U.S. Medicare beneficiaries at BIDMC used to test AI-DSA tool.

Sources:

  • "New Artificial Intelligence Tool Successfully Detects Severe Aortic Stenosis" (Beth Israel Deaconess Medical Center, September 26, 2024)
  • JACC: Advances, manuscript to be published (authors: Jordan B. Strom, MD, MSc, et al.)
  • National Echocardiographic Database of Australia
  • Beth Israel Deaconess Medical Center (BIDMC)
  • Beth Israel Lahey Health

Read more