New method spots first signs of cardiac arrhythmias
Cardiac arrhythmias kill over seven million people every year. The earlier they can be diagnosed and treated, the better. A recent issue of Nature’s Scientific Reports features a new state-of-the-art method for detecting the early warning signs of arrhythmias.
We all have an individual heartbeat. It is possible to apply machine learning methods to model all the potential arrhythmic beats of a healthy person and thereby identify irregular heartbeats as soon as they occur.
Cardiac arrhythmias count among the leading causes of death worldwide. In addition to genetic susceptibility, the risk of developing an irregular heartbeat increases, among others, with high blood pressure, diabetes, smoking, excessive consumption of alcohol, and stress. Arrhythmias often occur randomly at first. An international team of researchers from Tampere University of Technology, Qatar University and Izmir University of Economics has pioneered a new method for detecting the first signs of arrhythmias.
Conventional methods for predicting arrhythmias have proven unreliable. The new, patented method is currently 99.4 per cent accurate.
“If people have no history of heart problems, we can only analyse their normal heart rhythm. How can we know what their abnormal rhythm would look like? Our research found that it is possible to apply machine learning methods to model all the potential arrhythmic heartbeats of a healthy person,” says Professor Serkan Kiranyaz from Qatar University. He was a part of the team of researchers along with Professor Moncef Gabbouj from TUT and Turker Ince from Izmir University of Economics.
The team analysed extensive heart rate and arrhythmia datasets collected during their earlier ECG research.
“Our system is personalized for each user. A healthy user’s heart rate data and the synthesized abnormal heartbeats are entered into the system to serve as a baseline. This way the system is trained to monitor the user's heart rate and identify irregular heartbeats as soon as they occur,” adds Kiranyaz.
Consumer version on the horizon
Professor Moncef Gabbouj (left) from the Laboratory of Signal Processing and Qatar University Professor Serkan Kiranyaz, a frequent visitor on the TUT campus, have a long history of collaboration.
The system can be integrated into an affordable, portable device, similarly to wrist-worn activity bands or health trackers.
“These conventional devices may measure and monitor heart rates but cannot detect anomalies,” says Serkan Kiranyaz.
“We’re currently talking with several companies about the possibility of bringing the system to market. I’m confident that once we find the right company, product launch won’t be far off.”
The researchers assure that the system is extremely easy to use. Users upload their data to the server, and models of potential anomalies are transferred from the server to a hand-held device. The device alerts the user to any abnormal heart rhythms as soon as they occur.
“An added benefit is that the data is stored in the server and can be analysed later on with a cardiologist,” says Kiranyaz.