- My News
- All News
- Most Popular
Study data to be presented at the Heart Rhythm Society's (HRS) annual meeting suggest that a novel machine learning algorithm developed by Implicity significantly reduced the rate of false positives with Medtronic's insertable cardiac monitoring (ICM) devices. This was achieved "while maintaining detection of abnormalities (>99% sensitivity) and preserving patient safety," the authors said, adding their findings have "significant implications for clinic resources allocated to ICM review."
Investigators evaluated the IM007 algorithm, which adjudicates arrhythmias, in ICM electrocardiogram (ECG) strips from Medtronic Reveal devices. The study included 455 patients implanted with a Reveal LNQ11 ICM who were followed via the Implicity remote monitoring platform.
For each patient and each type of arrhythmia detected, one episode was randomly selected and analysed by the algorithm using the same labels of the LNQ11: asystole, bradycardia, atrial tachycardia/atrial fibrillation, ventricular tachycardia, artefact and normal. These ECG strips were selected from ICM episodes with durations of 9.5 seconds to five minutes and flagged as abnormal by the ICM. The same episodes were also independently adjudicated by a team of electrophysiologists, and the results compared to test the algorithm's performance.
Among 635 ICM episodes, there were 219, or just over 34%, adjudicated as ICM false positives, with the remaining 416 episodes classified as abnormal. Researchers reported that IM007 reclassified 174 of the 219 false positives as normal, while all but three of the 416 abnormal episodes remained that way. "All three false-negative episodes missed were adjudicated as regular [atrial tachycardia] associated with noise and considered as normal by at least one expert," the authors said. Overall, the results represent an absolute decrease of 25% of false positive episodes by using the algorithm, the study found.
To ensure you don't miss other Top Stories like these and news on key healthtech industry developments, sign up for our free daily e-newsletter here.
Did you like this article?