Prediction of Drug -Induced TdP Risks Using Machine Learning and Rabbit Ventricular Wedge Assay Jaela Foster -Burns a Nan Miles Xi b

2025-05-02 0 0 451.62KB 17 页 10玖币

侵权投诉

Prediction of Drug-Induced TdP Risks Using Machine Learning and Rabbit

Ventricular Wedge Assay

Jaela Foster-Burns a, Nan Miles Xi b,*

a Department of Biology, Loyola University Chicago, Chicago, IL 60660, USA

b Department of Mathematics and Statistics, Loyola University Chicago, Chicago, IL 60660, USA

* Correspondence: Nan Miles Xi (mxi1@luc.edu)

ABSTRACT

Torsades de pointes (TdP) is an irregular heart rhythm as a side effect of drugs and may cause

sudden cardiac death. A machine learning model that can accurately identify drug TdP risk is

necessary. This study uses multinomial logistic regression models to predict three-class drug TdP

risks based on datasets generated from rabbit ventricular wedge assay experiments. The training-

test split and five-fold cross-validation provide unbiased measurements for prediction accuracy.

We utilize bootstrap to construct a 95% confidence interval for prediction accuracy. The model

interpretation is further demonstrated by permutation predictor importance. Our study offers an

interpretable modeling method suitable for drug TdP risk prediction. Our method can be easily

generalized to broader applications of drug side effect assessment.

Keywords: torsades de pointes; drug safety assessment; machine learning; multinomial logistic

regression

INTRODUCTION

Torsades de Pointes (TdP) is a fatal polymorphic ventricular tachycardia. It is distinguished by

the elevated beating of the heart’s lower chambers (ventricles) and QT prolongation 1,2. The risk

of TdP may be signified on an electrocardiogram through a display of oscillatory changes in

amplitude of the QRS complexes around the isoelectric line. Experimental studies and literature

indicate that pharmaceutical drugs known to treat cardiomyopathies (heart disease) fall into

different risk levels of inducing TdP 3. In 2005 The International Committee/Council for

Harmonisation (ICH) established international regulatory guidelines, ICH S7B and ICH E14, for

the pharmaceutical industry to analyze drug TdP risk 4. Subsequentially, in 2013 the US Food and

Drug Administration (FDA) proposed the Comprehensive In Vitro Proarrhythmic Assay (CiPA)

initiative for improved drug-induced TdP risk prediction 5. A cumulation of these guidelines in

combination with in vitro (in the test tube, not in the organism) protocols has been utilized to

determine drug TdP risk within preclinical and clinical studies. These protocols use multielectrode

array or voltage-sensing optical approaches, combined with logistical ordinal and multinomial

linear regression models, to analyze electrophysiologic responses to drugs commonly linked to

low, medium, or high TdP risk categories 2,3.

The goals of this study surround the usage of data generated from in vitro preclinical studies

to predict drug TdP risk, along with the examination of variables contributing to drug TdP risk.

This study further examines the uncertainty of TdP drug risk prediction and identifies important

variables in the prediction. To achieve these goals, a multinomial logistic regression model,

training-test split, cross-validation, and permutation predictor importance are used in this study.

The drug TdP risk prediction is a three-classification problem that can be modeled with

multinomial random variables. The data for the model was acquired from the literature on a rabbit

ventricular wedge experiment 6.

The main result of this study indicates the risk classifications of 28 drugs known to induce TdP.

The risk classifications are determined by in vitro preclinical studies and are mainly supported by

ECG readings. The model places any drug into a low-, medium-, or high-risk category and reports

on the accuracy of the risk prediction while giving prediction uncertainty and a 95% confidence

interval. The significance of this study is that a model is trained to predict unknown data with

extremely high accuracy while stating the most important predictor. This study also outlines tools

to improve accuracy (cross-validation and bootstrap) or better the model.

DATASETS

This study utilizes data pulled from the literature on the utility of a normalized TdP score

system in drug proarrhythmic potential 6. The data in this literature originates from

experimentation done on rabbit ventricular wedge assays (RVWAs). The experimentation consists

of electrophysiological recordings conducted on surgically prepared rabbit left ventricular wedges

and is further detailed in supported literature 7. From experimentation, a transmural-pseudo-

electrocardiogram (ECG) is recorded, allowing for the calculation of TdP scores for each drug and

the development of TdP risk categories – low, medium, or high risk of TdP. This study uses the

findings from the electrocardiogram recordings to generate 15 variables to build the model

introduced (Table 1). Four replicates on each of the 28 drugs tested with RVWAs produce 112

observations in total for the model to be built upon (Table 2).

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

PredictionofDrug-InducedTdPRisksUsingMachineLearningandRabbitVentricularWedgeAssayJaelaFoster-Burnsa,NanMilesXib,*aDepartmentofBiology,LoyolaUniversityChicago,Chicago,IL60660,USAbDepartmentofMathematicsandStatistics,LoyolaUniversityChicago,Chicago,IL60660,USA*Correspondence:NanMilesXi(mxi1@luc.edu)A...

展开>> 收起<<

Prediction of Drug -Induced TdP Risks Using Machine Learning and Rabbit Ventricular Wedge Assay Jaela Foster -Burns a Nan Miles Xi b.pdf

共17页,预览4页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Prediction of Drug -Induced TdP Risks Using Machine Learning and Rabbit Ventricular Wedge Assay Jaela Foster -Burns a Nan Miles Xi b

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: