Cedars-Sinai, Stanford scientists use AI to flag hard-to-detect heart disease

Researchers from the Smidt Heart Institute at Cedars-Sinai have created an artificial intelligence tool to help identify hypertrophic cardiomyopathy and cardiac amyloidosis.  

Their findings, outlined in a study published in JAMA Cardiology, could make it easier to detect and treat heart disease.  

“These two heart conditions are challenging for even expert cardiologists to accurately identify, and so patients often go on for years to decades before receiving a correct diagnosis,” said Dr. David Ouyang, cardiologist in the Smidt Heart Institute and senior author of the study, in a statement.  

“Our AI algorithm can pinpoint disease patterns that can’t be seen by the naked eye, and then use these patterns to predict the right diagnosis,” he said.  


The study, which was also authored by researchers from Stanford University’s Department of Computer Science and the Imperial College London, sought to examine whether deep learning can be used to automate measurements of left ventricular dimensions, thereby flagging patients who might benefit from extra screening.  

Authors trained and tested the algorithm using independent echocardiogram videos from Stanford, Cedars-Sinai and the Unity Imaging Collaborative, an open-access data set from ICL.  

When applied to more than 34,000 clinical images, the algorithm identified features related to the thickness of heart walls and the size of heart chambers in order to draw attention to certain patients as potentially having the cardiac diseases.  

“The algorithm identified high-risk patients with more accuracy than the well-trained eye of a clinical expert,” said Ouyang.   

“This is because the algorithm picks up subtle cues on ultrasound videos that distinguish between heart conditions that can often look very similar to more benign conditions, as well as to each other, on initial review,” he continued.  

Hypertrophic cardiomyopathy – which causes the heart muscle to thicken and stiffen – and cardiac amyloidosis – caused by deposits of an abnormal protein in the heart tissue – can look very similar to each other on an ECG.   

However, the treatments differ dramatically.   

Additionally, as the researchers noted, both diseases can also look like a non-diseased heart in the early stages.  

Alongside the code for their algorithm and data-processing workflow, the scientists released their data set of 12,000 de-identified ECG videos, dramatically widening the availability of images for future AI approaches.   

Still, they acknowledged limitations, including potential bias in the available training images from patients.  

“For example, although hereditary cardiac amyloidosis is known to disproportionately affect Black individuals in the U.S. they are underrepresented in study cohorts, and care must be taken to extrapolate performance of deep learning algorithms in populations with different demographic characteristics,” they said.

Additionally, they noted that the model was trained on videos obtained by expert sonographers at an academic medical center.  

“With expansion in the use of point-of-care ultrasonography for evaluation of cardiac function by non-cardiologists, further work is needed to understand model performance with input videos of more variable quality and acquisition expertise as well as in comparison with other imaging modalities,” they wrote in the study.  


Researchers have made exciting strides over the years in the field of diagnostic AI. In addition to heart disease and heart failure, models have also been used to detect COVID-19 and lung cancer.  

However, there are potential downsides. In addition to potential bias, a recent study showed that diagnostic AI may be vulnerable to cyberattacks.  


“The use of artificial intelligence in cardiology has evolved rapidly and dramatically in a relatively short period of time,” said Dr. Susan Cheng, director of the Institute for Research on Healthy Aging in the Department of Cardiology at the Smidt Heart Institute and co-senior author of the study, in a statement.   

“These remarkable strides – which span research and clinical care – can make a tremendous impact in the lives of our patients,” she said.

Kat Jercich is senior editor of Healthcare IT News.
Twitter: @kjercich
Email: [email protected]
Healthcare IT News is a HIMSS Media publication.

Source: Read Full Article