Image: A new study asserts that AI can help identify silent heart disease (Photo courtesy of Shutterstock).
A new study reveals that artificial intelligence (AI) can successfully identify patients with congenital long QT syndrome (LQTS), despite having a normal QT interval on the standard electrocardiogram (ECG).
Researchers at the Mayo Clinic (Rochester, MN, USA) and AliveCor (San Francisco, CA, USA) conducted a retrospective study of 1,048 patients suffering from LQTS and a further 1,010 who were dismissed as such who had an ECG taken at the Mayo Genetic Heart Rhythm Clinic. An AI deep neural network (DNN) involving a multilayer convolutional recurrent neural network (CRNN) was then used to classify patients using a 10 second ECG from lead 1 alone; 72% of the ECGs were used for training and 28% for validation.
The results revealed that the DNN employed in the study generated an area under the curve of 0.83, with a specificity of 81%, sensitivity of 73%, and an overall accuracy of 79%. It was able to distinguish between the three main genotypes (LQTS1, LQTS2, and LQTS3). In addition, the DNN could successfully identify patients with congenital LQTS despite having a normal QT interval on a standard ECG, which is the case in as many as 50% of patients. The study was presented at the 40th annual Heart Rhythm Scientific Sessions conference, held during May 2018 in Boston (MA, USA).
"Although long QT syndrome is a potentially lethal syndrome, when it is recognized and treated, sudden death should almost never happen," said senior author Michael Ackerman, MD, PhD, director of Mayo Clinic's LQTS/Genetic Heart Rhythm Clinic. “The expectation needs to shift from merely preventing sudden death to enabling these patients and their families to live and thrive despite the diagnosis. Hopefully, the results of this study should be reassuring and encouraging to these families who live with long QT syndrome.”
LQTS is a condition that affects abnormal repolarization of the heart after a heartbeat, resulting in an increased risk fainting, drowning, or sudden death. It may be present at birth or develop later in life as a result of medication, low blood potassium, low blood calcium, or heart failure (HF). Medications that are implicated include certain anti-arrhythmic, antibiotics, and antipsychotics. Diagnosis is based on an ECG finding of a corrected QT interval of greater than 440-500 milliseconds and clinical findings. For people with LQTS who survive cardiac arrest and remain untreated, the risk of death within 15 years is greater than 50%, but with proper treatment, this decreases to less than 1% over 20 years.