An Overview of Machine Learning and Big Data for Drug Toxicity Evaluation

Abstract

Drug toxicity evaluation is an essential process of drug development as it is reportedly responsible for the attrition of approximately 30% of drug candidates. The rapid increase in the number and types of large toxicology data sets together with the advances in computational methods may be used to improve many steps in drug safety evaluation. The development of in silico models to screen and understand mechanisms of drug toxicity may be particularly beneficial in the early stages of drug development where early toxicity assessment can most reduce expenses and labor time. To facilitate this, machine learning methods have been employed to evaluate drug toxicity but are often limited by small and less diverse data sets. Recent advances in machine learning methods together with the rapid increase in big toxicity data such as molecular descriptors, toxicogenomics, and high-throughput bioactivity data may help alleviate some of the current challenges. In this article, the most common machine learning methods used in toxicity assessment are reviewed together with examples of toxicity studies that have used machine learning methodology. Furthermore, a comprehensive overview of the different types of toxicity tools and data sets available to build in silico toxicity prediction models has been provided to give an overview of the current big toxicity data landscape and highlight opportunities and challenges related to them.

Assessing the Calibration in Toxicological in Vitro Models with Conformal Prediction

“…2 Alzheimer's Research UK UCL Drug Discovery Institute, WC1E 6BT London, United Kingdom. 3 Department of Pharmaceutical Biosciences and Science for Life Laboratory, Uppsala University, 751 24 Uppsala, Sweden. 4 Division of Computational Science and Technology, KTH, 100 44 Stockholm, Sweden.…”

“…Machine Learning (ML) methods are ubiquitous in drug discovery and toxicity prediction [1,2]. In silico toxicity prediction is typically used to guide toxicity testing in early phases of drug design [3]. With more high-quality standardised data available, the (potential) impact of ML methods in regulatory toxicology is growing.…”

Human Induced Pluripotent Stem Cells as a Screening Platform for Drug-Induced Vascular Toxicity

“…CRISPR-based gene editing tools together with patient-specific iPSCs may unveil the effects of single nucleotide polymorphism (SNP) on drug response and hence determine individual predisposition to toxicity (Pang et al, 2009;Burridge et al, 2016;Seeger et al, 2017;Garg et al, 2018;Ma et al, 2018). Machine learning supported with big data has also shown promise in improving the prediction of drug toxicity (Lau et al, 2019;Vo et al, 2020). Collectively, these technologies have led us to uncharted territory for toxicology.…”

Transcriptomics in Toxicogenomics, Part I: Experimental Design, Technologies, Publicly Available Data, and Regulatory Aspects

“…For this reason, one exposure time point results in only revealing a small fraction of the whole biological outcome. Additionally, enough dose and time points need to be measured in order to predict long-term toxic effects through advanced computational approaches and machine learning algorithms [41,42], as also discussed in detail in the third part of this review. Nonetheless, most of the gene expression changes are measurable in hours or days after exposure, rather than months or years.…”