Machine learning approach to evaluate TdP risk of drugs using cardiac electrophysiological model including inter-individual variability

Aug 5, 2026

Upload by Anda

A multi-in silico framework predicts drug-induced Torsade de Pointes (TdP) risk by combining electrophysiological modeling with machine learning. A virtual population of human ventricular cell models, built from a modified O’Hara-Rudy model to represent inter-individual variability, is simulated using IC50 and Hill coefficients from 67 drugs. Fourteen simulation-derived features train KNN, Random Forest, XGBoost, and ANN models with grid-search optimization and five-fold cross-validation. The ANN achieves the strongest performance on unseen drug data, supporting reliable TdP risk assessment.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages18

Reading Time45 min

AccessFree to download

Introduction

Drug-induced TdP and existing risk guidelines

Method

Virtual population modeling and input drug parameters

Feature extraction and machine learning models

Result

Predictive performance of the proposed ANN model

Discussion and conclusion

Generalization on unseen datasets

TYPE Original Research PUBLISHED 04 October 2023 DOI 10.3389/fphys.2023.1266084 OPEN ACCESS EDITED BY Morten Gram Pedersen, University of Padua, Italy REVIEWED BY Francisco Sahli Costabal, Pontiﬁcia Universidad Católica de Chile, Chile Frederique Jos Vanheusden, Nottingham Trent University, United Kingdom *CORRESPONDENCE Ki Moo Lim, [kmlim@kumoh.ac.kr](kmlim@kumoh.ac.kr) RECEIVED 24 July 2023 ACCEPTED 20 September 2023 PUBLISHED 04 October 2023 CITATION Fuadah YN, Qauli AI, Marcellinus A, Pramudito MA and Lim KM (2023), Machine learning approach to evaluate TdP risk of drugs using cardiac electrophysiological model including inter-individual variability. Front. Physiol. 14:1266084 . doi: 10.3389/fphys.2023.1266084 COPYRIGHT © 2023 Fuadah, Qauli, Marcellinus, Pramudito and Lim. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Machine learning approach to evaluate TdP risk of drugs using cardiac electrophysiological model including inter-individual variability Yunendah Nur Fuadah 1,2, Ali Ikhsanul Qauli 1,3, Aroli Marcellinus 1, Muhammad Adnan Pramudito 1 and Ki Moo Lim 1,4,5* 1Computational Medicine Lab, Department of IT Convergence Engineering, Kumoh National Institute of Technology, Gumi, Republic of Korea, 2School of Electrical Engineering, Telkom University, Bandung, Indonesia, 3Department of Engineering, Faculty of Advanced Technology and Multidiscipline, Universitas Airlangga, Surabaya, Jawa Timur, Indonesia, 4Computational Medicine Lab, Department of Medical IT Convergence Engineering, Kumoh National Institute of Technology, Gumi, Republic of Korea, 5Meta Heart Co., Ltd., Gumi, Republic of Korea Introduction: Predicting ventricular arrhythmia Torsade de Pointes (TdP) caused by drug-induced cardiotoxicity is essential in drug development. Several studies used single biomarkers such as qNet and Repolarization Abnormality (RA) in a single cardiac cell model to evaluate TdP risk. However, a single biomarker may not encompass the full range of factors contributing to TdP risk, leading to divergent TdP risk prediction outcomes, mainly when evaluated using unseen data. We addressed this issue by utilizing multi-in silico features from a population of human ventricular cell models that could capture a representation of the underlying mechanisms contributing to TdP risk to provide a more reliable assessment of druginduced cardiotoxicity. Method: We generated a virtual population of human ventricular cell models using a modiﬁed O ’ Hara-Rudy model, allowing inter-individual variation. IC50 and Hill coefﬁcients from 67 drugs were used as input to simulate drug effects on cardiac cells. Fourteen features (~~d~~V~~d~~ repol , ~~d~~V~~d~~ max , Vmpeak , Vmresting , APDtri , APD90 , APD50 , Capeak , Cadiastole , Catri , CaD90 , CaD50 , qNet, qInward) could be generated from the simulation and used as input to several machine learning models, including k-nearest neighbor (KNN), Random Forest (RF), XGBoost, and Artiﬁcial Neural Networks (ANN) . Optimization of the machine learning model was performed using a grid search to select the best parameter of the proposed model. We applied ﬁve-fold cross-validation while training the model with 42 drugs and evaluated the model ’ s performance with test data from 25 drugs. Result: The proposed ANN model showed the highest performance in predicting the TdP risk of drugs by providing an accuracy of 0.923 (0.908–0.937), sensitivity of 0.926 (0.909–0.942), speciﬁcity of 0.921 (0.906–0.935), and AUC score of 0. 964 (0 .954–0. 975) . Discussion and conclus

FAQs

What modeling approach is used to represent inter-individual variability in drug effects?

A virtual population of human ventricular cell models is generated using a modified O’Hara-Rudy model, enabling inter-individual variation during electrophysiological simulation.

Which machine learning methods are evaluated in the study?

The study evaluates KNN, Random Forest, XGBoost, and Artificial Neural Networks (ANN), using simulation-derived features as inputs.

How is model performance assessed and optimized?

Model parameters are selected via grid search, training uses five-fold cross-validation with 42 drugs, and performance is evaluated on test data from 25 unseen drugs.

Aug 5, 2026

Upload by Anda