Deep learning analysis of electrocardiogram for risk prediction of drug-induced arrhythmias and diagnosis of long QT syndrome
importance 6/10
ai assisted
Cardiac electrophysiology and arrhythmias cardiac drug safety stage: research
Problem
QTc measurement has low positive predictive value for identifying drugs at high risk of torsade de pointes and diagnosing congenital long QT syndrome.
Why it matters
TdP is a life-threatening ventricular arrhythmia, so better risk prediction could prevent drug-induced deaths.
What the AI did
CNN models quantify ECG alterations induced by sotalol to enhance prediction of drug-induced TdP and diagnosis of cLQTS.
Results
not stated in source
Impact
Models were tested on cohorts of 1029 healthy subjects, 487 cLQTS patients, and 48 diTdP patients.
Record
| Venue | European Heart Journal |
|---|---|
| Published | 2021 |
| Work type | article |
| Authors | Edi Prifti, Ahmad Fall, Giovanni Davogustto, Alfredo Pulini, Isabelle Denjoy, Christian Funck‐Brentano, Yasmin West Khan, Alexandre Durand-Salmon et al. |
| Citations | 83 (OpenAlex, at import time) |
| References | 45 |
| Model or method | convolutional neural network (CNN) models |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1093/eurheartj/ehab588 |
| Open access | hybrid |
| Keywords | ECG, deep learning, long QT syndrome, torsade de pointes, arrhythmia risk |
Importance rating
6
significant — thousands to millions affected
Abstract calls TdP a life-threatening arrhythmia; better prediction than low-PPV QTc could improve drug-safety screening and cLQTS diagnosis.