International Validation of Echocardiographic AI Amyloid Detection Algorithm
2024

AI Algorithm for Detecting Cardiac Amyloidosis

Sample size: 1423 publication Evidence: high

Author Information

Author(s): Duffy Grant BS, Oikonomou Evan MD, DPhil, Hourmozdi Jonathan MD, Usuku Hiroki MD, PhD, Patel Jigesh MD, Stern Lily MD, Goto Shinichi MD, PhD, Tsujita Kenichi MD, PhD, Khera Rohan MD, MS, Ahmad Faraz S. MD, Ouyang David MD

Primary Institution: Smidt Heart Institute, Cedars-Sinai Medical Center, Los Angeles, CA, USA

Hypothesis

Can a computer vision detection algorithm accurately identify cardiac amyloidosis in echocardiogram studies across multiple international sites?

Conclusion

The EchoNet-LVH algorithm can assist with earlier and accurate diagnosis of cardiac amyloidosis.

Supporting Evidence

  • EchoNet-LVH had an AUC of 0.896.
  • At the pre-specified threshold, EchoNet-LVH had a sensitivity of 0.644 and specificity of 0.988.
  • The positive predictive value was 0.968 and negative predictive value was 0.828.
  • Performance was consistent across different sites and patient demographics.

Takeaway

A smart computer program can help doctors find a rare heart disease called cardiac amyloidosis more quickly and accurately.

Methodology

A multi-site retrospective case-control study evaluating the performance of the EchoNet-LVH algorithm using echocardiogram videos.

Potential Biases

There is potential observer variability in echocardiographic measurements.

Limitations

The study is retrospective and lacks prospective trialing of CA screening approaches.

Participant Demographics

Mean age of participants was 78.2 years, with 77.7% male.

Statistical Information

Confidence Interval

95% CI 0.875 – 0.916

Statistical Significance

p<0.05

Digital Object Identifier (DOI)

10.1101/2024.12.14.24319049

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication