AI-ENHANCED Detection of Aortic Stenosis

Presented at the ESC congress 2022 by: Prof. Geoffrey Strange, MD, PhD- Sydney, Australia

Introduction and methods

Although aortic valve replacement (AVR) is associated with a large reduction in mortality risk in patients with aortic stenosis (AS), AS remains largely undertreated.

This study investigated whether an AI-Decision Support Algorithm (AI-DSA) developed from echocardiographic parameters could identify moderate-to-severe and severe AS. The study used data from the National Echo Database of Australia (NEDA), which contains data from 1,077,145 routine echocardiograms from 631,824 patients from 23 centers in Australia with 7.2±4.4 years maximal follow-up, and is linked to the national deaths index in Australia. 442276 echocardiograms (70%) were used to train the AI model, and 189548 echocardiograms (30%) were used as for a validation test to assess the performance of the model.

Main results

• 5-year mortality per AI-DSA output was 22.9% in the low probability group (reference group), 56.2% in patients with moderate-to-severe AS (OR 1.82, 95% CI 1.63-2.02, P<0.001) and 67.9% in those with severe AS (OR 2.80, 95% CI 2.57-3.06, P<0.001).
• Within the group identified as severe AS by AI-DSA, those that met current guidelines for severe AS had a 5-year mortality of 69.1% (reference). Those who were identified by AI-DSA as severe AS, but who do not meet current guidelines, had a 5-year-mortality rate of 64.4% (OR 1.26, 95% CI 1.04-1.53, P=0.021).

Conclusion

This study showed that an AI algorithm could automatically identify patients with moderate-to-severe and severe forms of AS associated with a high mortality risk.

-Our reporting is based on the information provided at the ESC Congress-