New Artificial Intelligence System Can Help Predict Cardiac Arrests

To enhance the accuracy of predicting cardiac arrest, researchers from Johns Hopkins University created an artificial intelligence (AI) system based on raw images of patient hearts and data on their demographics.

Known as Survival Study of Cardiac Arrhythmia Risk (SSCAR), the AI system can detect patterns that cannot be seen with the naked eye and thereby can predict the probability of sudden cardiac arrest over the following 10 years.

"The images carry critical information that doctors haven't been able to access," said study first author Dan Popescu, a former Johns Hopkins doctoral student, in the press release. "This scarring can be distributed in different ways and it says something about a patient's chance for survival. There is information hidden in it." 

The AI system collects more information than volume and mass, which are the only features that a standard clinical cardiac image analysis collects.

In addition to this, the team created another neural network that used 10 years of patient data which included 22 different patient factors, including age, weight, and race.

Researchers used patient data from 60 health centers across the country to validate the two algorithms. The various types of imaging data and cases that the algorithms used to make predictions imply their ability to function anywhere.

"This has the potential to significantly shape clinical decision-making regarding arrhythmia risk and represents an essential step towards bringing patient trajectory prognostication into the age of artificial intelligence," said senior author Natalia Trayanova, co-director of Johns Hopkins' Alliance for Cardiovascular Diagnostic and Treatment Innovation, in a press release. "It epitomizes the trend of merging artificial intelligence, engineering, and medicine as the future of healthcare." 

Further, Johns Hopkins researchers hope to create similar AI algorithms for various other heart diseases.

Using AI systems to assist in detecting and managing cardiovascular diseases is becoming increasingly common.

Recently, physicians from the Society of Nuclear Medicine and Molecular Imaging found that analyzing coronary 18F-NaF uptake on PET and quantitative coronary plaque characteristics via an AI model showed many signs of heart attack risk.

Another AI approach created by a Mayo Clinic research team was able to use voice biomarkers to define any potential heart issues. The approach worked by providing all patients with access to a smartphone app where they entered three 30-second voice recordings. Analyzing elements such as pitch, cadence, and frequency, the app used the recordings to locate any clogged arteries.

A study from December 2019 also showed how machine learning tools can be used to predict a heart attack. Researchers collected data on coronary artery calcium using non-contrast computed tomography and then used machine learning to predict the likelihood of a heart attack. Researchers determined that the predictions of the tools were accurate.