Cardiac Arrest

How about knowing about Cardiac Arrest in Advance? Check it out!

Healthcare Healthcare IT Pharmaceuticals

The healthcare vertical is ushering a survival predictor, the first-of-its-kind AI-based approach, which is capable of predicting as to when would someone succumb to cardiac arrest, that too, in a precise manner.

The technology mentioned above has been built based on raw images of the diseased hearts along with the background of patients. As such, the healthcare personnel could work on increasing the survival of patients from lethal and sudden cardiac arrhythmias. Details have been published in “Nature Cardiovascular Research”, dated April 7, 2022 by researchers from Johns Hopkins University.


According to Natalia Trayanova, a senior author and Murray B. Sachs professor of Biomedical Engineering and Medicine, abrupt cardiac death caused due to arrhythmia contributes for more than 20% of fatalities across the globe. The professor further stated that the scenario such that patients at a lower risk of abrupt cardiac death receiving defibrillators and those at a higher risk of this type of death not receiving these defibrillators on time is indeed glaring. The AI-based algorithm could gauge the ones needing this treatment on an urgent basis. This would, in turn, help the healthcare personnel to attend to them on priority, thereby mitigating the further complications.

The team is reported to make use of neural networks for building personalized survival assessment for every heart patient. These risk measures do make provisions for higher-degree precision regarding foretelling the chance of cardiac arrest (and ultimately, death) over the subsequent decade.

Deep Learning Technology

The above-mentioned technology is termed as “deep learning technology”, better known as SSCAR (Survival Study of Cardiac Arrhythmia Risk). This name does refer to cardiac scarring resulting out of heart disease that generally culminates into lethal arrhythmias. This is actually the key to various predictions to algorithm.

The researchers have used contrast-enhanced cardiac images visualizing scar distribution through loads of real patients at Johns Hopkins Hospital for training the algorithm for detecting relationships and patterns that are not visible to naked eyes. The existing cardiac image analysis does extract merely simple scar features such as mass and volume, thereby underutilizing critical data contained in the images.

First author and former Johns Hopkins doctoral student “Dan Popescu” affirms that this scarring could be distributed in various ways and it speaks volumes about chances of survival of a patient. This information could be extracted for good using the algorithm.

The 2nd neural network was trained for learning from 10 years of prescribed clinical patient data. There were 22 factors like age, weight, prescription drug use, and race.

60 health centers in the US were considered for these trainings. This would suffice the ongoing trend of merging engineering, AI, and medicine. This would, in fact, be the future of healthcare.

The researchers are now into building algorithms for detecting the other cardiac ailments. Trayanova states that this concept of deep learning could be extended for the other streams of medicine relying on virtual diagnosis.

So, the department of cardiology is bound to see deep learning –based turbulence in the next 10 years.

Leave a Reply

Your email address will not be published.