AI used to detect ‘invisible’ heart condition

Investigators from the Mayo Clinic and AliveCor demonstrated that an AI network can help identify ‘invisible’ heart condition

Investigators from the Mayo Clinic and AliveCor demonstrated that a trained artificial intelligence network can help identify people with increased risks of arrhythmias and sudden cardiac death, despite displaying a normal heart rhythm on their electrocardiogram.

Up to half of patients with long QT syndrome can show a normal interval on a standard test, the personal EKG manufacturer AliveCor say in a statement. Correct diagnoses and treatment can be crucial, especially when using drugs that may prolong heartbeats.

The researchers’ deep neural network generated the results using data from a single lead of a 12-lead EKG—measuring the voltage between the left and right arms—suggesting that simpler, portable devices may be used to detect the concealed heart condition, the company announced. The network had an overall accuracy of 79%, with 73% sensitivity and 81% specificity.

“There can be no better illustration of the importance of our AI to medical science than using it to detect that which is otherwise invisible,” says AliveCor CEO Vic Gundotra. A study abstract was unveiled at Heart Rhythm 2018, the Heart Rhythm Society’s 39th Annual Scientific Sessions (9 May–12 May, Boston, USA).

The inherited form of long QT syndrome affects 160,000 people in the USA, causing 3,000 to 4,000 deaths in children and young adults annually. LQTS can also be caused by nearly 100 FDA-approved medications, including certain antibiotics and antidepressants, AliveCor say.


Please enter your comment!
Please enter your name here