Noninvasive assessment of dofetilide plasma concentration using a deep learning (neural network) analysis of the surface electrocardiogram: A proof of concept study

• DOI: 10.1371/journal.pone.0201059
• Publication Date: 2018-08-22T13:27:17Z
• Authors: Zachi I. Attia; Alan Sugrue; Samuel J. Asirvatham; Michael J. Ackerman; Suraj Kapa; Paul A. Friedman; Peter A. Noseworthy
• Keywords: Dofetilide

Dofetilide is an effective antiarrhythmic medication for rhythm control in atrial fibrillation, but carries a significant risk of pro-arrhythmia and requires meticulous dosing monitoring. The cornerstone this monitoring, measurement the QT/QTc interval, imperfect surrogate plasma concentration, efficacy, pro-arrhythmic potential.The aim our study was to test application deep learning approach (using convolutional neural network) assess morphological changes on surface ECG (beyond QT interval) relation dofetilide concentrations.We obtained publically available serial ECGs drug concentrations from 42 healthy subjects who received or placebo placebo-controlled cross-over randomized controlled clinical trial. Three replicate 10-s were extracted at predefined time-points with simultaneous concentration We developed algorithm predict 30 then tested model remaining 12 subjects. compared leaning linear based only QTc.Fourty two (21 females, 21 males) studied mean age 26.9 ± 5.5 years. A QTc correlated reasonably well levels (r = 0.64). best correlation level achieved 0.85).This proof concept suggests that artificial intelligence (deep learning/neural applied superior analysis interval alone predicting concentration.