AI Detect Heart Failure More Accurately Than Doctors

Vaibhav Kapadia - Oct 06, 2019


AI Detect Heart Failure More Accurately Than Doctors

Researchers used AI to develop a method to detect heart failure with 100% accuracy through analysis of just a raw electrocardiogram signal.

About 26 million people worldwide are currently affected by a form of heart failure. Researchers have recently used AI (Artificial Intelligence) to develop a method to detect heart failure. They can determine whether a person has CHF (congestive heart failure) or not with 100% accuracy, by analyzing only a raw ECG (electrocardiogram) signal.

Picture1
AI can detect heart failure more accurately than doctors

CHF is a condition that occurs when your heart muscle stops pumping blood through your body as well as it should. There are many people has CHF. It also has high health care costs and a pretty high mortality rate. Therefore, health systems and clinical practitioners require finding effective methods of detection.

In order to solve these problems, the researchers used a hierarchical neural network called CNN (short for Convolutional Neural Networks), which are highly effective in recognizing structures and patterns in data. Their research was published in the Biomedical Signal Processing and Control Journal.

Picture2
Researchers used Convolutional Neural Networks to detect CHF

The researchers tested their CNN model on large ECG data sets, which contained subjects with CHF and also healthy hearts. They used a combination of machine learning and advanced signal processing tools on raw ECG signals. The results showed that this model can detect CHF with 100% accuracy through analysis of just a heartbeat.

Study researcher Leandro Pecchia from the University of Warwick said that their research shows a major improvement over current CHF detection methods. These methods often based on heart rate variability to detect CHF, which are effective, but error-prone and time-consuming. The CNN model in the new study, meanwhile, can deliver 100% accuracy by checking just a raw ECG signal.

Picture3
The CNN model can detect CHF with 100% accuracy through analysis of a heartbeat

According to Sebastiano Massaro, one of the study researchers and also an associate professor at the University of Surrey, this is one of the first models that can identify the ECG's morphological features that specifically related to the severity of CHF.

Tags

0 Comments

Sort by Newest | Popular

Next Story