A new intelligence-based technique developed by the University of Johns Hopkins researchers can be reported to predict if and when a patient can die due to a heart attack. According to the university, technology trained in raw images of liver and painful patient backgrounds increase doctor’s predictions and have the potential to revolutionize clinical decision making and increase survival opportunities for patients with sudden and deadly cardiac arrhythmias.
The researchers documented their findings in a paper entitled “Prediction of sudden survival of arrhythmias using a deep analysis of learning from scarring in the liver” published in natural cardiovascular research.”Sudden heart death caused by arrhythmia accounts for 20% of all deaths around the world and we know a little about why it happens or how to say who is at risk,” said Natalia Trayanova, a senior writer on a university press statement, in a statement. Trayanoya is a professor of biomedical engineering and medicine at the university.
“There are patients who may be at low risk of sudden heart death which makes the defibrillators that they might not need and then there are high-risk patients who do not get the care they need and can die in their lives. What our algorithm can do is determine who is at risk of experiencing Heart death and when it will occur, allowing doctors to decide what needs to be done, “explained TRayanoya.The research team uses neural networks to build personsomeized survival assessments to each patient with heart disease. This assessment is reported to be able to predict the chances of sudden heart death more than 10 years, and even when it is most likely to occur.
In-depth learning technology named the study of survival on the cardiac arrhythmia (SSCAR) as a figure of scar heart caused by heart disease, which often leads to arrhythmias. The researchers used a heart image that was enhanced in contrast from hundreds of real patients (with a heart scar tape) at Johns Hopkins Hospital to train algorithms to detect patterns and relationships that are not visible to the naked eye.According to universities, the current clinical heart image analysis only extracts simple scar features such as volume and mass. This means that they greatly reduce what new algorithms are shown to be important data.
The researchers also train second neural networks to learn from 10 years of standard data that cover 22 factors such as the age of patients, weight, race and use of prescription drugs.Universities report that algorithm predictions are significantly more accurate in each size than a doctor and that they are validated in tests with separate and independent patient cohorts of 60 health centers throughout the United States.According to Trayanya, this in-depth learning concept can be developed for other fields of drugs that rely on visual diagnosis.