by John R. Fischer
, Senior Reporter | April 28, 2022
A new algorithm developed at Johns Hopkins University may indicate if and when a patient will die from cardiac arrest.
Leveraging information from raw MR images, the AI approach is significantly more accurate than a human doctor at determining the odds of sudden cardiac death over 10 years, and when it’s most likely to happen.
Called Survival Study of Cardiac Arrhythmia Risk (SSCAR), the application visualizes and measures scar tissue. It is the first combination of neural networks that provides a personalized survival assessment for each patient with heart disease.
For those who need to move fast and expand clinical capabilities -- and would love new equipment -- the uCT 550 Advance offers a new fully configured 80-slice CT in up to 2 weeks with routine maintenance and parts and Software Upgrades for Life™ included.
“Sudden cardiac death caused by arrhythmia accounts for as many as 20% of all deaths worldwide, and we know little about why it’s happening or how to tell who’s at risk. There are patients who may be at low risk of sudden cardiac death getting defibrillators that they might not need, and then there are high-risk patients that aren’t getting the treatment they need and could die in the prime of their life,” said senior author Natalia Trayanova, the Murray B. Sachs professor of biomedical engineering and medicine at JHU, in a statement.
Current clinical cardiac image analysis only measures volume and mass of scar tissue and does not take into account other critical data. Trayanova and her colleagues trained their algorithm using contrast-enhanced cardiac images that show scar distribution from 156 patients at Johns Hopkins Hospital. This allows it to identify patterns and relationships not observable to the naked eye.
They trained a second neural network to study 10 years of standard clinical patient data and 22 factors such as patient age, weight, race and prescription drug use. A survival model then used information from both neural networks to make predictions and estimate the level of uncertainty with each result.
When retrospectively applied to 113 patients with coronary heart disease, the algorithm was more accurate on every measure in its predictions than doctors. The independent patient cohort came from 60 health centers across the U.S. with different cardiac histories and imaging data. This indicates that the platform may be usable anywhere.
Trayanova, who is also the co-director of the Alliance for Cardiovascular Diagnostic and Treatment Innovation, and the rest of the team are building algorithms to detect other cardiac diseases, based on the same deep-learning concept. She says it could also be used for visual diagnosis of non-cardiac-related diseases. “This has the potential to significantly shape clinical decision-making regarding arrhythmia risk and represents an essential step toward bringing patient trajectory prognostication into the age of artificial intelligence.”
The findings were published in Nature Cardiovascular Research
JHU did not respond for comment.