About 10 percent of adults who are over the age of 65 suffer from some type of congestive heart failure (CHF). There are a variety of causes for congestive heart failure, but the chronic condition commonly results from the heart being unable to pump blood effectively through the body.
Blood tests, x-rays, and ultrasounds all provide clinicians useful ways to diagnose CHF, but more commonly used methods involve using an electrocardiogram (ECG) signals to figure out heart rate variability in few minutes, or even multiple measurements over the days depending on the requirement. An impressive new technology has been demonstrated, using a convolutional neural network (CNN) that can detect CHF instantly by checking ECG data from a single heartbeat.
Sebastian Massaro from the University of Surrey explained that the researchers trained and tested the CNN model on large and publicly available ECG datasets examining subjects with CHF as well as healthy arrhythmic hearts. The model delivered 100 percent accuracy. The model showed promise as the researchers were able to detect whether or not the person has heart failure by checking just one heartbeat. The model is one of the first known to identify the ECG’s morphological features specifically associated with the severity of the condition.
How promising is the new technology?
According to Massaro and his team, the model currently delivers an astonishing 100 percent accuracy rate, but the research also has some limitations. The data used in the current study only consisted of ECG readings from extreme cases of CHF and healthy patients. The researchers are still not sure whether it would provide 100 percent accuracy when tested on patients with mild CHF. Therefore more work is required to verify a broader spectrum of CHF diagnosis before the technology hits clinical practices.
The latest technology carves a path for the development of several exciting AI-driven diagnostic tools currently being used in the medical field or is under development, which promises a revolution in clinical approaches to evaluating medical data. Recently, a team from the Mayo Clinic trained a neural network to identify the patients suffering from asymptomatic left ventricular dysfunction, a forerunner to heart failure that is difficult for clinicians to detect, using just 10 seconds of ECG data.
The technology will provide the possibility of wearable health monitoring devices being able to help doctors to identify at-risk patients without having to examine them in clinical contexts. Sebastian Massaro and his team suggest that their work, using short ECG recordings to detect CHF, could build a path for health wearables that monitor patients with real-world conditions.
The researchers also believe that this is an important result as with the increasing availability of wearable devices capturing interim ECG recordings, like smartwatches, accurate CHF detection and prediction might soon be performed through devices used by people regularly.