AI Accurately Detects Heart Failure Using Just One Heartbeat
New technological discoveries are constantly improving the way doctors and medical professionals do their jobs. With approximately 550,000 new cases of heart failure diagnosed each year in the US alone, docs need all the help they can get.
A recent study published in Biomedical Signal Processing and Control Journal determined that doctors can now detect heart failure with 100% accuracy from just a single heartbeat using a new artificial intelligence-driven neural network. This study was conducted to explore how new technology can improve previous methods of detecting congestive heart failure.
The study was led by a group of researchers from the University of Surrey, Warwick, and Florence. They determined that AI can accurately identify CHF by analyzing one ECG heartbeat. Heathline defines congestive heart failure (CHF) as a chronic progressive condition that affects the pumping power of your heart's muscles. Research has shown that 5 million people live with CHF in the United States.
Researchers from the study believe that clinical practitioners and health systems “urgently require efficient detection processes” because of "high prevalence, significant mortality rates and sustained healthcare costs".
The researchers hope that they can solve this by using convolutional neural networks (CNN). CNN is more reliable for identifying patterns and structures of data.
Doctors previously used time-consuming and inaccurate methods to detect CHF. This new AI method uses both advanced signal processing and machine learning tools on raw ECG signals to significantly improve detection rates.
“We trained and tested the CNN model on large publicly available ECG datasets featuring subjects with CHF as well as healthy, non-arrhythmic hearts. Our model delivered 100% accuracy: by checking just one heartbeat we are able to detect whether or not a person has heart failure. Our model is also one of the first known to be able to identify the ECG' s morphological features specifically associated to the severity of the condition,” said Dr. Massaro, associate professor of organizational neuroscience at the University of Surrey.
"First, by assessing ECG directly, we confirm that with AI it is possible to accurately detect CHF looking beyond heart rate variability analysis. Thus, we have in general results that are more adherent to the real behavior of the affected heart,” Dr. Massaro added.
A different part of the experiment used a specific CNN model to improve the accuracy of CHF detection while considering comparable models.
“We focus on the detection of the pathology from one single heartbeat in excerpts of 5-minutes rather than in 24-hours recordings,” said Dr. Massaro.
“This aspect offers a valuable potential for prospects of rapid interventions; nonetheless it is also important to keep in mind that we are talking about severe CHF patients only at the moment.”
In the near future, Dr. Massaro would like to expand the study and extend the approach to large scale samples, as well as, other classes of CHF. The goal for this is that it will eventually be implemented in everyday health care systems and practices.
Dr. Massaro added, “The application of organizational neuroscience, and specifically of neural network approaches to healthcare issues promises to open breakthrough frontiers for both clinical research and practice.”