Machine learning in drug development

Abstract All medications have adverse effects. Among the most serious of these are cardiac arrhythmias. Current paradigms for drug safety evaluation are costly, lengthy, conservative, and impede efficient drug development. Here we combine multiscale experiment and simulation, high-performance computing, and machine learning to create an easy-to-use risk assessment diagram to quickly and reliable stratify new and existing drugs according to their pro-arrhythmic potential. We capitalize on recent developments in machine learning and integrate information across ten orders of magnitude in space and time to provide a holistic picture of the effects of drugs, either individually or in combination with other drugs. We show, both experimentally and computationally, that drug-induced arrhythmias are dominated by the interplay between two currents with opposing effects: the rapid delayed rectifier potassium current and the L-type calcium current. Using Gaussian process classification, we create a classifier that stratifies drugs into safe and arrhythmic domains for any combinations of these two currents. We demonstrate that our classifier correctly identifies the risk categories of 23 common drugs, exclusively on the basis of their concentrations at 50% current block. Our new risk assessment diagram explains under which conditions blocking the L-type calcium current can delay or even entirely suppress arrhythmogenic events. Using machine learning in drug safety evaluation can provide a more accurate and comprehensive mechanistic assessment of the pro-arrhythmic potential of new drugs. Our study paves the way towards establishing science-based criteria to accelerate drug development, design safer drugs, and reduce heart rhythm disorders.

[1]  Jiang Yao,et al.  Predicting critical drug concentrations and torsadogenic risk using a multiscale exposure-response simulator. , 2019, Progress in biophysics and molecular biology.

[2]  F. S. Costabal,et al.  Machine learning in drug development: Characterizing the effect of 30 drugs on the QT interval using Gaussian process regression, sensitivity analysis, and uncertainty quantification. , 2019, Computer methods in applied mechanics and engineering.

[3]  Natalia Trayanova,et al.  From genetics to smart watches: developments in precision cardiology , 2018, Nature Reviews Cardiology.

[4]  N. Trayanova,et al.  Computational models in cardiology , 2018, Nature Reviews Cardiology.

[5]  David G Strauss,et al.  CiPA challenges and opportunities from a non-clinical, clinical and regulatory perspectives. An overview of the safety pharmacology scientific discussion. , 2018, Journal of pharmacological and toxicological methods.

[6]  George E Karniadakis,et al.  Omics, big data and machine learning as tools to propel understanding of biological mechanisms and to discover novel diagnostics and therapeutics. , 2018, Metabolism: clinical and experimental.

[7]  Jiang Yao,et al.  Predicting drug‐induced arrhythmias by multiscale modeling , 2018, International journal for numerical methods in biomedical engineering.

[8]  Francesca N. Delling,et al.  Heart Disease and Stroke Statistics—2018 Update: A Report From the American Heart Association , 2018, Circulation.

[9]  Weihua Li,et al.  In Silico Prediction of Chemical Toxicity for Drug Design Using Machine Learning Methods and Structural Alerts , 2018, Front. Chem..

[10]  E. Kuhl,et al.  Predicting the cardiac toxicity of drugs using a novel multiscale exposure–response simulator , 2018, Computer methods in biomechanics and biomedical engineering.

[11]  Min Wu,et al.  Uncertainty Quantification Reveals the Importance of Data Variability and Experimental Design Considerations for in Silico Proarrhythmia Risk Assessment , 2017, Front. Physiol..

[12]  Jaimit Parikh,et al.  Novel Two-Step Classifier for Torsades de Pointes Risk Stratification from Direct Features , 2017, Front. Pharmacol..

[13]  Gary R. Mirams,et al.  Early afterdepolarisation tendency as a simulated pro-arrhythmic risk indicator , 2017, bioRxiv.

[14]  A. Krekhov,et al.  Fast propagation regions cause self-sustained reentry in excitable media , 2017, Proceedings of the National Academy of Sciences.

[15]  Zhilin Qu,et al.  Stochastic initiation and termination of calcium-mediated triggered activity in cardiac myocytes , 2017, Proceedings of the National Academy of Sciences.

[16]  Gary R. Mirams,et al.  Hierarchical Bayesian inference for ion channel screening dose-response data , 2016, Wellcome open research.

[17]  E. Sobie,et al.  Improved Prediction of Drug‐Induced Torsades de Pointes Through Simulations of Dynamics and Machine Learning Algorithms , 2016, Clinical pharmacology and therapeutics.

[18]  Euan A Ashley,et al.  Early somatic mosaicism is a rare cause of long-QT syndrome , 2016, Proceedings of the National Academy of Sciences.

[19]  Zhihua Li,et al.  Optimization of an In silico Cardiac Cell Model for Proarrhythmia Risk Assessment , 2016, Front. Physiol..

[20]  T. Colatsky,et al.  The Comprehensive in Vitro Proarrhythmia Assay (CiPA) initiative - Update on progress. , 2016, Journal of pharmacological and toxicological methods.

[21]  D. Strauss,et al.  An evaluation of 30 clinical drugs against the comprehensive in vitro proarrhythmia assay (CiPA) proposed ion channel panel. , 2016, Journal of pharmacological and toxicological methods.

[22]  Ellen Kuhl,et al.  Generating Purkinje networks in the human heart. , 2016, Journal of biomechanics.

[23]  G. Gintant,et al.  Evolution of strategies to improve preclinical cardiac safety testing , 2016, Nature Reviews Drug Discovery.

[24]  T. Hisada,et al.  Screening system for drug-induced arrhythmogenic risk combining a patch clamp and heart simulator , 2015, Science Advances.

[25]  Ellen Kuhl,et al.  The Living Heart Project: A robust and integrative simulator for human heart function. , 2014, European journal of mechanics. A, Solids.

[26]  Edward W. Lowe,et al.  Computational Methods in Drug Discovery , 2014, Pharmacological Reviews.

[27]  S. Göktepe,et al.  Computational modeling of chemo‐electro‐mechanical coupling: A novel implicit monolithic finite element approach , 2013, International journal for numerical methods in biomedical engineering.

[28]  A. Garfinkel,et al.  Early afterdepolarizations in cardiac myocytes: beyond reduced repolarization reserve. , 2013, Cardiovascular research.

[29]  J. Verducci,et al.  MICE Models: Superior to the HERG Model in Predicting Torsade de Pointes , 2013, Scientific Reports.

[30]  N. Stockbridge,et al.  Dealing with Global Safety Issues , 2013, Drug Safety.

[31]  Gary R. Mirams,et al.  Computational assessment of drug-induced effects on the electrocardiogram: from ion channel to body surface potentials , 2013, British journal of pharmacology.

[32]  Gareth J Waldron,et al.  Reducing safety-related drug attrition: the use of in vitro pharmacological profiling , 2012, Nature Reviews Drug Discovery.

[33]  Yoram Rudy,et al.  Simulation of the Undiseased Human Cardiac Ventricular Action Potential: Model Formulation and Experimental Validation , 2011, PLoS Comput. Biol..

[34]  Gary R. Mirams,et al.  Simulation of multiple ion channel block provides improved early prediction of compounds’ clinical torsadogenic risk , 2011, Cardiovascular research.

[35]  A. Garfinkel,et al.  Early afterdepolarizations and cardiac arrhythmias. , 2010, Heart rhythm.

[36]  Robert B. Gramacy,et al.  Particle Learning of Gaussian Process Models for Sequential Design and Optimization , 2009, 0909.5262.

[37]  S. Göktepe,et al.  Computational modeling of cardiac electrophysiology: A novel finite element approach , 2009 .

[38]  D. Noble,et al.  Mathematical models of the electrical action potential of Purkinje fibre cells , 2009, Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences.

[39]  Diana X. Tran,et al.  Synchronization of chaotic early afterdepolarizations in the genesis of cardiac arrhythmias , 2009, Proceedings of the National Academy of Sciences.

[40]  Kousik Krishnan,et al.  Current concepts in the mechanisms and management of drug-induced QT prolongation and torsade de pointes. , 2007, American heart journal.

[41]  J. Valentin,et al.  A Rabbit Langendorff Heart Proarrhythmia Model: Predictive Value for Clinical Identification of Torsades de Pointes , 2006, British journal of pharmacology.

[42]  D. Noble,et al.  A model for human ventricular tissue. , 2004, American journal of physiology. Heart and circulatory physiology.

[43]  A. Camm,et al.  Relationships between preclinical cardiac electrophysiology, clinical QT interval prolongation and torsade de pointes for a broad range of drugs: evidence for a provisional safety margin in drug development. , 2003, Cardiovascular research.

[44]  R. W. Hansen,et al.  The price of innovation: new estimates of drug development costs. , 2003, Journal of health economics.

[45]  A. Garfinkel,et al.  Preventing ventricular fibrillation by flattening cardiac restitution. , 2000, Proceedings of the National Academy of Sciences of the United States of America.

[46]  A. Karma,et al.  New paradigm for drug therapies of cardiac fibrillation. , 2000, Proceedings of the National Academy of Sciences of the United States of America.

[47]  R Lazzara,et al.  Multiple mechanisms in the long-QT syndrome. Current knowledge, gaps, and future directions. The SADS Foundation Task Force on LQTS. , 1996, Circulation.

[48]  Craig T. January,et al.  Early Afterdepolarizations: Mechanism of Induction and Block A Role for L‐Type Ca2+ Current , 1989, Circulation research.

[49]  S. Polak,et al.  Am I or am I not proarrhythmic? Comparison of various classifications of drug TdP propensity. , 2017, Drug discovery today.