This recent AI heart disease detector can’t be beat.
Structural heart disease (SHD) refers to defects in the center’s valves, wall or chambers which might be present at birth or develop over time. These abnormalities can impair the center’s ability to pump blood effectively.
SHD is typically described as “hidden” heart disease because it could possibly progress without noticeable symptoms — until there’s a significant event like a heart attack or stroke.
Now, researchers at Columbia University and NewYork-Presbyterian have developed an AI-
powered screening tool to discover who should undergo a key ultrasound used to diagnose structural heart problems.
“There was a growth within the variety of AI models to detect, or opportunistically screen, disease,” Dr. Pierre Elias, an assistant professor of drugs and biomedical informatics at Columbia’s Vagelos College of Physicians and Surgeons, told The Post.
“A few of the most fun can look for coronary disease on CT scans or take a look at mammograms to assist doctors find breast cancer more accurately,” he added. “EchoNext is the primary model to detect all types of structural heart disease from ECGs.”
An electrocardiogram (ECG) is a fast, non-invasive procedure that measures the center’s electrical activity.
It’s one of the ceaselessly used cardiac tests, often ordered when patients experience symptoms resembling shortness of breath, chest pain, palpitations or sudden lack of consciousness.
While an ECG can detect some heart conditions, it’s not reliable for catching SHD by itself.
Enter EchoNext. The tool, fine-tuned over 4 years, analyzes ECG data to find out when follow-up with an echocardiogram is obligatory.
An echocardiogram is an ultrasound imaging test used to diagnose a variety of heart conditions, including valve disorders and congenital heart defects.
“EchoNext principally uses the cheaper test to work out who needs the costlier ultrasound,” said Elias, study leader and medical director for artificial intelligence at NewYork-Presbyterian.
“It detects diseases cardiologists can’t from an ECG,” he continued. “We expect that ECG plus AI has the potential to create a wholly recent screening paradigm.”
EchoNext was trained on over 1.2 million ECG–echocardiogram pairs from 230,000 patients.
The tool accurately detected 77% of structural heart problems on 3,200 ECGs, outperforming 13 cardiologists who logged a 64% accuracy.
EchoNext then identified over 7,500 people from a pool of nearly 85,000 study participants as high risk for undiagnosed SHD.
The researchers followed the patients for a yr without telling their physicians in regards to the forewarning.
Some 55% went on to have their first echocardiogram. Of those, almost three-quarters were diagnosed with SHD, a much higher positivity rate than usual.
The findings were published Wednesday within the journal Nature.
“The goal is to get the fitting patients to the fitting doctor and treatment sooner,” Elias said.
“The fact is many patients that need a cardiologist are sometimes missed, and EchoNext helps facilitate getting these patients to the cardiologist who can then get the patient to the treatment they need.”
Looking ahead, Columbia has submitted a patent application on the EchoNext ECG algorithm.
A clinical trial to check EchoNext in eight emergency departments can also be underway.
This recent AI heart disease detector can’t be beat.
Structural heart disease (SHD) refers to defects in the center’s valves, wall or chambers which might be present at birth or develop over time. These abnormalities can impair the center’s ability to pump blood effectively.
SHD is typically described as “hidden” heart disease because it could possibly progress without noticeable symptoms — until there’s a significant event like a heart attack or stroke.
Now, researchers at Columbia University and NewYork-Presbyterian have developed an AI-
powered screening tool to discover who should undergo a key ultrasound used to diagnose structural heart problems.
“There was a growth within the variety of AI models to detect, or opportunistically screen, disease,” Dr. Pierre Elias, an assistant professor of drugs and biomedical informatics at Columbia’s Vagelos College of Physicians and Surgeons, told The Post.
“A few of the most fun can look for coronary disease on CT scans or take a look at mammograms to assist doctors find breast cancer more accurately,” he added. “EchoNext is the primary model to detect all types of structural heart disease from ECGs.”
An electrocardiogram (ECG) is a fast, non-invasive procedure that measures the center’s electrical activity.
It’s one of the ceaselessly used cardiac tests, often ordered when patients experience symptoms resembling shortness of breath, chest pain, palpitations or sudden lack of consciousness.
While an ECG can detect some heart conditions, it’s not reliable for catching SHD by itself.
Enter EchoNext. The tool, fine-tuned over 4 years, analyzes ECG data to find out when follow-up with an echocardiogram is obligatory.
An echocardiogram is an ultrasound imaging test used to diagnose a variety of heart conditions, including valve disorders and congenital heart defects.
“EchoNext principally uses the cheaper test to work out who needs the costlier ultrasound,” said Elias, study leader and medical director for artificial intelligence at NewYork-Presbyterian.
“It detects diseases cardiologists can’t from an ECG,” he continued. “We expect that ECG plus AI has the potential to create a wholly recent screening paradigm.”
EchoNext was trained on over 1.2 million ECG–echocardiogram pairs from 230,000 patients.
The tool accurately detected 77% of structural heart problems on 3,200 ECGs, outperforming 13 cardiologists who logged a 64% accuracy.
EchoNext then identified over 7,500 people from a pool of nearly 85,000 study participants as high risk for undiagnosed SHD.
The researchers followed the patients for a yr without telling their physicians in regards to the forewarning.
Some 55% went on to have their first echocardiogram. Of those, almost three-quarters were diagnosed with SHD, a much higher positivity rate than usual.
The findings were published Wednesday within the journal Nature.
“The goal is to get the fitting patients to the fitting doctor and treatment sooner,” Elias said.
“The fact is many patients that need a cardiologist are sometimes missed, and EchoNext helps facilitate getting these patients to the cardiologist who can then get the patient to the treatment they need.”
Looking ahead, Columbia has submitted a patent application on the EchoNext ECG algorithm.
A clinical trial to check EchoNext in eight emergency departments can also be underway.
