Key result
CNN using in silico action potentials accurately classifies drug TdP risk with ~0.91 high-risk AUC.
Why the study?
Machine-learning features automatically extracted from action potential shapes were hypothesized to contain more meaningful information for proarrhythmic risk assessment than manually extracted indicators.
Does a convolutional neural network (CNN) model using raw in silico action potential shapes accurately classify the proarrhythmic Torsade de Pointes (TdP) risk of drugs into high-, intermediate-, and low-risk categories?
Does a convolutional neural network (CNN) model using raw in silico action potential shapes accurately classify the proarrhythmic Torsade de Pointes (TdP) risk of drugs into high-, intermediate-, and low-risk categories?
A convolutional neural network trained directly on raw in silico action potential shapes can accurately classify high- and low-risk torsadogenic drugs without manual feature engineering, offering a potential preclinical screening tool for drug cardiotoxicity.
This study proposes a convolutional neural network (CNN) model using action potential (AP) shapes as input for proarrhythmic risk assessment, considering the hypothesis that machine-learning features automatically extracted from AP shapes contain more meaningful information than do manually extracted indicators. We used 28 drugs listed in the comprehensive in vitro proarrhythmia assay (CiPA), consisting of eight high-risk, eleven intermediate-risk, and nine low-risk torsadogenic drugs. We performed drug simulations to generate AP shapes using experimental drug data, obtaining 2000 AP shapes per drug. The proposed CNN model was trained to classify the TdP risk into three levels, high-, intermediate-, and low-risk, based on in silico AP shapes generated using 12 drugs. We then evaluated the performance of the proposed model for 16 drugs. The classification accuracy of the proposed CNN model was excellent for high- and low-risk drugs, with AUCs of 0.914 and 0.951, respectively. The model performance for intermediate-risk drugs was good, at 0.814. Our proposed model can accurately assess the TdP risks of drugs from in silico AP shapes, reflecting the pharmacokinetics of ionic currents. We need to secure more drugs for future studies to improve the TdP-risk-assessment robustness.
No takes yet. Share an insight, caveat, or question.
Jeong et al. (2023) studied Drug-induced torsade de pointes (TdP) proarrhythmic risk (n=28). Convolutional neural network (CNN) model using in silico action potential (AP) shapes vs. Reference CiPA proarrhythmic risk categories (high-, intermediate-, and low-risk) was evaluated on Classification accuracy (AUC) for predicting drug-induced torsade de pointes (TdP) risk categories (high-, intermediate-, and low-risk) in the 16-drug test set. A convolutional neural network using in silico action potential shapes classified drug TdP risk with AUCs of 0.914 for high-risk, 0.814 for intermediate-risk, and 0.951 for low-risk drugs.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:
More from Biomedicines