Study says AI reads heart studies better than sonographers

Share This Post

- Advertisement -

Artificial intelligence (AI) has been shown to be superior in assessing and diagnosing heart function compared to echocardiographic evaluations performed by imaging technicians, according to research in which 3,495 tests were evaluated.

- Advertisement -

The results of the randomized, blinded clinical trial are published in nature magazinein an article led by experts at Cedars-Sinai Medical Center in Los Angeles, who wanted to answer the question of who can best evaluate and diagnose heart function after reading an echocardiogram: AI or a sonographer?

In 2020, researchers from the Smidt Heart Institute, part of Cedars-Sinai, together with Stanford University, developed an AI model to evaluate heart functionin particular, the left ventricular ejection fraction.

- Advertisement -
Doctors could not distinguish which assessments were made by AI and which by sonographers, Photo Shutterstock.

Doctors could not distinguish which assessments were made by AI and which by sonographers, Photo Shutterstock.

This is a key measure used in the diagnosis of cardiac function; his description was also published at the time in Nature.

Based on these findings, the published study examines whether or not AI is more accurate in scoring tests.

According to the results, cardiologists more frequently agreed with the initial assessment of AI and made corrections only in 16.8% of the first evaluations made by this technology.

However, cardiologists corrected 27.2% of the initial assessments made by sonographerssays a statement from Cedars-Sinai.

According to the same source, the doctors were unable to distinguish which assessments were performed by the AI ​​and which by sonographers, and IThe help of artificial intelligence has saved timeOR.

“We asked our cardiologists to guess whether the preliminary interpretation was done by AI or a sonographer, and it turned out they couldn’t tell the difference,” says David Ouyang.

Cardiologists more frequently agreed with the initial AI assessment.  Shutterstock photos.

Cardiologists more frequently agreed with the initial assessment of AI. Shutterstock photos.

“This speaks to the strong performance of the AI ​​algorithm, as well as the seamless integration into the clinical software. We believe all of this are good signals for future research of trials with AI in this field”.

His hope, he says, “is to save clinicians time and minimize the most tedious parts of the cardiac imaging workflow. However, the cardiologist remains the expert who finally decides on the results of the AI ​​model”.

The work, according to Ouyang, provides “rigorous evidence that using artificial intelligence in this new way can improve the quality and efficacy of echocardiogram images for many patients.”

“This successful clinical trial sets a great precedent for how new clinical AI algorithms can be discovered and tested within healthcare systems, increasing the likelihood of continued implementation to improve patient care.” concludes Sumeet Chugh. .

EFE Agency

Source: Clarin

- Advertisement -

Related Posts