FRIDAY, March 23, 2018 (HealthDay News) -- A smartwatch coupled with a machine learning algorithm is able to accurately detect atrial fibrillation (AF), with some loss of specificity and sensitivity compared to criterion-standard electrocardiography (ECG), according to a study published online March 21 in JAMA Cardiology.
Geoffrey H. Tison, M.D., M.P.H., from the University of California, San Francisco, and colleagues used heart rate and step count data obtained from 9,750 participants enrolled in the Health eHeart Study. Data were used to develop an algorithm capable of detecting AF on a smartwatch.
The researchers trained the algorithm using 139 million heart rate measurements. The deep neural network exhibited a C statistic of 0.97 (P < 0.001) to detect AF against the reference standard of AF diagnosed with a 12-lead ECG. In the external validation cohort of 51 patients undergoing cardioversion, the algorithm's sensitivity and specificity were 98.0 and 90.2 percent, respectively. Using self-report of persistent AF in ambulatory participants for exploratory analysis, the C statistic was 0.72, sensitivity was 67.7 percent, and specificity was 67.6 percent.
"This proof-of-concept study found that smartwatch photoplethysmography coupled with a deep neural network can passively detect AF but with some loss of sensitivity and specificity against a criterion-standard ECG," the authors write.
Several authors disclosed financial ties to medical device companies, including Cardiogram, which partially funded the study.
Abstract/Full Text (subscription or payment may be required)
Editorial (subscription or payment may be required)
