Combining machine learning models of in vitro and in vivo bioassays improves rat carcinogenicity prediction

Guan, Davy; Fan, Kevin; Spence, Ian; Matthews, Slade

doi:10.1016/j.yrtph.2018.01.008

Cited by 20 publications

(13 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On our dataset the decision trees classifier, somewhat surprisingly, had better results than random forests, although in other cases, the latter has shown superior performance [42]. SVM and AdaBoost classifiers, which for other datasets have been very successful [43][44][45][46], in our case, did not perform as well. The balancing techniques (downsampling, naïve random oversampling and SMOTE) generally contributed little to the improvement of performance, particularly in models built with the different fingerprints.…”

Section: Discussionmentioning

confidence: 59%

QSAR Models for Active Substances against Pseudomonas aeruginosa Using Disk-Diffusion Test Data

2021

View full text Add to dashboard Cite

Pseudomonas aeruginosa is a Gram-negative bacillus included among the six “ESKAPE” microbial species with an outstanding ability to “escape” currently used antibiotics and developing new antibiotics against it is of the highest priority. Whereas minimum inhibitory concentration (MIC) values against Pseudomonas aeruginosa have been used previously for QSAR model development, disk diffusion results (inhibition zones) have not been apparently used for this purpose in the literature and we decided to explore their use in this sense. We developed multiple QSAR methods using several machine learning algorithms (support vector classifier, K nearest neighbors, random forest classifier, decision tree classifier, AdaBoost classifier, logistic regression and naïve Bayes classifier). We used four sets of molecular descriptors and fingerprints and three different methods of data balancing, together with the “native” data set. In total, 32 models were built for each set of descriptors or fingerprint and balancing method, of which 28 were selected and stacked to create meta-models. In terms of balanced accuracy, the best performance was provided by KNN, logistic regression and decision tree classifier, but the ensemble method had slightly superior results in nested cross-validation.

show abstract

Section: Discussionmentioning

confidence: 59%

QSAR Models for Active Substances against Pseudomonas aeruginosa Using Disk-Diffusion Test Data

2021

View full text Add to dashboard Cite

show abstract

“…It has been most widely used to solve binary classification problems. , SVM can also be used to predict toxicity end points, such as estrogen receptor activity, hepatocyte toxicity, and inflammation. ,, kNN is a nonparametric method that classifies targets to be predicted based on the nearest training data in a feature space . kNN has been used to predict acute toxicity (LD50), chronic toxicity, genotoxicity, carcinogenicity, neurotoxicity, and androgen receptor activity. ,− NB is a simple probabilistic classifier based on Bayesian rules and is used for conditional probabilities . It has been used to predict reproductive toxicity, aquatic toxicity, developmental toxicity, and mitochondrial toxicity. ,,, Finally, a decision tree (DT) analyzes data to create patterns as a combination of predictable rules .…”

Section: Model Algorithmsmentioning

confidence: 99%

Artificial Intelligence-Based Toxicity Prediction of Environmental Chemicals: Future Directions for Chemical Management Applications

Jeong

Choi

2022

Environ. Sci. Technol.

View full text Add to dashboard Cite

Recently, research on the development of artificial intelligence (AI)-based computational toxicology models that predict toxicity without the use of animal testing has emerged because of the rapid development of computer technology. Various computational toxicology techniques that predict toxicity based on the structure of chemical substances are gaining attention, including the quantitative structure–activity relationship. To understand the recent development of these models, we analyzed the databases, molecular descriptors, fingerprints, and algorithms considered in recent studies. Based on a selection of 96 papers published since 2014, we found that AI models have been developed to predict approximately 30 different toxicity end points using more than 20 toxicity databases. For model development, molecular access system and extended-connectivity fingerprints are the most commonly used molecular descriptors. The most used algorithm among the machine learning techniques is the random forest, while the most used algorithm among the deep learning techniques is a deep neural network. The use of AI technology in the development of toxicity prediction models is a new concept that will aid in achieving a scientific accord and meet regulatory applications. The comprehensive overview provided in this study will provide a useful guide for the further development and application of toxicity prediction models.

show abstract

“…Wang et al 84 developed a new DL architecture, CapsCarcino, to distinguish between carcinogens and noncarcinogens, achieving an accuracy of 85% on an external validation data set. Guan et al 85 developed and combined QSAR models with a variety of ML algorithms for the prediction of carcinogenic properties and compared results with existing software. Their model predicted rat carcinogenicity with 69% accuracy and 70% AUC.…”

Section: Ai‐based Toxicity Predictionmentioning

confidence: 99%

Toxicity prediction based on artificial intelligence: A multidisciplinary overview

Pérez‐Santín

Rodríguez‐Solana

García

et al. 2021

WIREs Comput Mol Sci

View full text Add to dashboard Cite

The use and production of chemical compounds are subjected to strong legislative pressure. Chemical toxicity and adverse effects derived from exposure to chemicals are key regulatory aspects for a multitude of industries, such as chemical, pharmaceutical, or food, due to direct harm to humans, animals, plants, or the environment. Simultaneously, there are growing demands on the authorities to replace traditional in vivo toxicity tests carried out on laboratory animals (e.g., European Union REACH/3R principles, Tox21 and ToxCast by the U.S. government, etc.) with in silica computational models. This is not only for ethical aspects, but also because of its greater economic and time efficiency, as well as more recently because of their superior reliability and robustness than in vivo tests, mainly since the entry into the scene of artificial intelligence (AI)‐based models, promoting and setting the necessary requirements that these new in silico methodologies must meet. This review offers a multidisciplinary overview of the state of the art in the application of AI‐based methodologies for the fulfillment of regulatory‐related toxicological issues. This article is categorized under: Data Science > Chemoinformatics Data Science > Artificial Intelligence/Machine Learning

show abstract

Combining machine learning models of in vitro and in vivo bioassays improves rat carcinogenicity prediction

Cited by 20 publications

References 45 publications

QSAR Models for Active Substances against Pseudomonas aeruginosa Using Disk-Diffusion Test Data

QSAR Models for Active Substances against Pseudomonas aeruginosa Using Disk-Diffusion Test Data

Artificial Intelligence-Based Toxicity Prediction of Environmental Chemicals: Future Directions for Chemical Management Applications

Toxicity prediction based on artificial intelligence: A multidisciplinary overview

Contact Info

Product

Resources

About