Scientists show how AI finds invisible signs of heart failure

Researchers at Mount Sinai School of Medicine have developed an electrocardiogram reading algorithm that can detect subtle signs of heart failure. Credits: Glicksberg and Nadkarni labs, Mount Sinai, NY, NY

A special artificial intelligence (AI) -based computer algorithm created by researchers at Mount Sinai identifies subtle changes in electrocardiography (also known as ECG or EKG) to determine if a patient is experiencing heart failure. I was able to learn how to predict.

“We have shown that deep learning algorithms can recognize problems with blood pumps on both sides. heart From the ECG waveform data. “Dr. Benjamin S. Gricksburg, Assistant Professor of Genetics and Genomics, Member of Hassoplatner Digital Health Institute on Mount Sinai, and Journal of the American College of Cardiology: Cardiovascular Imaging.. “Usually, diagnosing these types of heart conditions requires costly and time-consuming procedures. This algorithm allows heart failure.. “

The study includes postdoctoral researchers Akhil Vaid and MD working at the Glicksberg Lab, and Girish N, an associate professor at the Icahn School of Medicine at Mount Sinai School of Medicine. It was led by Nadkarni, MD, MPH and CPH. Senior author of data-driven and digital medicine (D3M) departments and research.

Heart failure, or congestive heart failure, which affects about 6.2 million Americans, occurs when the heart pumps less blood than the body normally needs. For years, doctors have relied heavily on imaging techniques called echocardiography to assess whether a patient is experiencing heart failure. Echocardiography is useful, but it can be a labor-intensive procedure offered only in certain hospitals.

However, recent advances in artificial intelligence suggest that the widely used electrical recorder, the electrocardiogram, may be a fast and readily available alternative in these cases. For example, many studies have shown how “deep learning” algorithms can detect weaknesses in the heart. Left ventricle, Pushes freshly oxygenated blood to the rest of the body. In this study, researchers described the development of an algorithm that assesses the strength of the right ventricle as well as the left ventricle. This algorithm takes in the oxygenated blood that flows in from the body and sends it to the lungs.

“Although attractive, it has traditionally been difficult for physicians to use ECG to diagnose heart failure. This is due to the lack of established diagnostic criteria for these assessments and the ECG reading. Some changes in the values ​​are so subtle that they cannot be detected by the human eye. “Dr. Nadkarni said,” This study uses relatively simple and widely available tests to better screen and It represents an exciting step in finding hidden information in ECG data that can lead to a therapeutic paradigm. “

An electrocardiogram usually involves a two-step process. The leads are taped to various parts of the patient’s chest, and a specially designed portable machine prints a series of wavy lines or waveforms that represent the electrical activity of the heart within minutes. These machines are found in most hospitals and ambulances throughout the United States and require minimal training to operate.

In this study, researchers programmed a computer to read a patient’s electrocardiogram, along with data extracted from a written report summarizing the results of the corresponding echocardiograms taken from the same patient. In this situation, the written report served as a standard dataset for computers to compare with electrocardiographic data and learn how to find weak hearts.

Natural language processing programs helped computers extract data from written reports. Meanwhile, a special neural network has been incorporated that can discover patterns in the image, allowing the algorithm to learn to recognize the strength of the pump.

“We wanted to drive cutting-edge technology by developing AI that could easily and inexpensively understand the entire heart,” said Dr. Vaid.

The computer then read more than 700,000 electrocardiographic and echocardiographic reports from 150,000 Mount Sinai health system patients between 2003 and 2020. We trained the computer using data from four hospitals and tested the performance of the algorithm using data from a fifth hospital. In a different experimental setting.

“The potential advantage of this study is that it contains one of the largest collections of ECGs from one of the world’s most diverse patient populations,” said Dr. Nadkarni.

Early results suggested that the algorithm was effective in predicting which patients had a healthy or very weak left ventricle.Here the strength was defined by the left ventricle Ejection fraction, An estimate of the amount of fluid drained by the ventricles at each beat, as observed on echocardiography. A healthy heart has an ejection fraction of 50% or more, but a weak heart has an ejection fraction of 40% or less.

The algorithm was 94% accurate in predicting which patients had a healthy ejection fraction and 87% accurate in predicting patients with an ejection fraction of less than 40%.

However, the algorithm was not very effective in predicting which patients had a slight heart weakness. In this case, the program was 73% accurate in predicting patients with ejection fractions of 40-50%.

Further results suggested that the algorithm also learned to detect weaknesses in the right valve from the electrocardiogram. In this case, the weaknesses were defined by more descriptive terms extracted from the echocardiographic report. Here, the algorithm was 84 percent accurate in predicting which patients had a weak right valve.

“Our results suggest that this algorithm may ultimately help physicians correctly diagnose disorders on both sides of the heart,” said Dr. Vaid.

Finally, additional analysis suggests that this algorithm may be effective in detecting heart weakness in all patients, regardless of race or gender.

“Our result is that this is algorithm It could be a useful tool for clinicians to help various patients fight heart failure suffering, “added Dr. Gricksburg.

The AI ​​algorithm is consistent with the expertise of a cardiologist, explaining that decision

For more information:
Akhil Vaid et al, Use of Deep Learning Algorithms to Simultaneously Identify Right and Left Ventricular Dysfunctions from Electrocardiograms, JACC: Cardiovascular imaging (2021). DOI: 10.1016 / j.jcmg.2021.08.004

Quote: Scientists Get Invisible Signs of Heart Failure (October 19, 2021) from Shows how

This document is subject to copyright. No part may be reproduced without written permission, except for fair transactions for personal investigation or research purposes. The content is provided for informational purposes only.

Scientists show how AI finds invisible signs of heart failure

Source link Scientists show how AI finds invisible signs of heart failure

Show More

Related Articles

Back to top button