Mixed learning algorithms and features ensemble in hepatotoxicity prediction

Liew, Chin Yee; Lim, Yen Ching; Yap, Chun Wei

doi:10.1007/s10822-011-9468-3

Cited by 84 publications

(103 citation statements)

References 75 publications

(82 reference statements)

Supporting

Mentioning

102

Contrasting

Order By: Relevance

“…Nonetheless, from the view of the performance of DL-NCTR model, the DL method showed a powerful learning ability relevant to DILI prediction. 8 Therefore, when using the basically same datasets for training and prediction, the performance of the DL-Liew model was confirmed to be more powerful for DILI prediction than that of the model constructed by Liew et al. 8…”

Section: Dl-nctr Dili Modelmentioning

confidence: 86%

Deep Learning for Drug-Induced Liver Injury

Dai

Chen

et al. 2015

J. Chem. Inf. Model.

299

249

View full text Add to dashboard Cite

Drug-induced liver injury (DILI) has been the single most frequent cause of safety-related drug marketing withdrawals for the past 50 years. Recently, deep learning (DL) has been successfully applied in many fields due to its exceptional and automatic learning ability. In this study, DILI prediction models were developed using DL architectures, and the best model trained on 475 drugs predicted an external validation set of 198 drugs with an accuracy of 86.9%, sensitivity of 82.5%, specificity of 92.9%, and area under the curve of 0.955, which is better than the performance of previously described DILI prediction models. Furthermore, with deep analysis, we also identified important molecular features that are related to DILI. Such DL models could improve the prediction of DILI risk in humans. The DL DILI prediction models are freely available at http://www.repharma.cn/DILIserver/DILI_home.php.

show abstract

Section: Dl-nctr Dili Modelmentioning

confidence: 86%

Deep Learning for Drug-Induced Liver Injury

Dai

Chen

et al. 2015

J. Chem. Inf. Model.

299

249

View full text Add to dashboard Cite

show abstract

“…Nevertheless, the maximum ensemble size tested in this study was limited to only 101 which may be insufficient for stacking to work properly. As observed in a previous ensemble study (ESMdesc) for hepatotoxicity, [29] the stacking method requires at least 400 base models to obtain acceptable performance.…”

Section: Effects Of Training Set Sampling Methodsmentioning

confidence: 96%

“…To generate ensemble models of varied training set and features (ESM train + desc ), the process (with descriptor grouping) outlined in a previous study on ensemble of varied features (ESM desc ) [29] was adapted with an additional step to sample training data at the start. The process is illustrated in Figure 2.…”

Section: Ensemble Methodsmentioning

confidence: 99%

“…Seven descriptor groups were used in our method: 2DMisc, Chi, Counts, EStateCount, EStateSum, MDE, and MLFER. [29] We have used the strategy of overproduce and select, [32] where a large pool of base classifiers was generated, but only a few selected models will be combined into an ensemble model. To obtain base classifiers, briefly: 1.…”

Section: Ensemble Methodsmentioning

confidence: 99%

See 1 more Smart Citation

QSAR and Predictors of Eye and Skin Effects

Liew

Yap

2013

Molecular Informatics

Self Cite

View full text Add to dashboard Cite

In this study, the ensemble of features and training samples was examined with a collection of support vector machines. The effects of data sampling methods, ratio of positive to negative compounds, and types of base models combiner to produce ensemble models were explored. The ensemble method was applied to produce four separate in silico models to classify the labels for eye/skin corrosion (H314), skin irritation (H315), serious eye damage (H318), and eye irritation (H319), which are defined in the "Globally Harmonized System of Classification and Labelling of Chemicals". To the best of our knowledge, the training set used in this work is one of the largest (made of publicly available data) with acceptable prediction performances. These models were distributed via PaDEL-DDPredictor (http://padel.nus.edu.sg/software/padelddpredictor) that can be downloaded freely for public use.

show abstract

“…There is therefore a key need for models that better predict drug-induced liver injury in humans. Furthermore, models that also more reliably predict drug metabolization, pharmacodynamics, pharmacokinetics, and pharmacogenomics are a cornerstone of pharmaceutical development (Liew et al, 2011;Watkins, 2011).…”

Section: Introductionmentioning

confidence: 99%

An Ex Vivo Perfusion System Emulating In Vivo Conditions in Noncirrhotic and Cirrhotic Human Liver

Schreiter

Marquitan

Darnell

et al. 2012

J Pharmacol Exp Ther

View full text Add to dashboard Cite

Various models are used for investigating human liver diseases and testing new drugs. However, data generated in such models have only limited relevance for in vivo conditions in humans. We present here an ex vivo perfusion system using human liver samples that enables the characterization of parameters in a functionally intact tissue context. Resected samples of noncirrhotic liver (NC; n ϭ 10) and cirrhotic liver (CL; n ϭ 12) were perfused for 6-h periods. General and liver-specific parameters (glucose, lactate, oxygen, albumin, urea, and bile acids), liver enzymes (aspartate aminotransferase, alanine aminotransferase, lactate dehydrogenase, glutamate dehydrogenase, and ␥-glutamyl transferase), overall (M65) and apoptotic (M30) celldeath markers, and indicators of phase-I/phase-II biotransformations were analyzed. The measurement readings closely resembled (patho)physiological characteristics in patients with NC and CL. Mean courses of glucose levels reflected the CLs' reduced glycogen storage capability. Furthermore, CL samples exhibited significantly stronger increases in lactate, bile acids, and the M30/M65 ratio than NC specimens. Likewise, NC samples exhibited more rapid phase-I transformations of phenacetin, midazolam, and diclofenac and phase-I to phase-II turnover rates of the respective intermediates than CL tissue. Collectively, these findings reveal the better hepatic functionality in NC. Perfusion of human liver tissue with this system emulates in vivo conditions and clearly discriminates between noncirrhotic and cirrhotic tissue. This highly reliable device for investigating basic hepatic functionality and testing safety/toxicity, pharmacokinetics/pharmacodynamics and efficacies of novel therapeutic modalities promises to generate superior data compared with those obtained via existing economic perfusion systems.

show abstract

Mixed learning algorithms and features ensemble in hepatotoxicity prediction

Cited by 84 publications

References 75 publications

Deep Learning for Drug-Induced Liver Injury

Deep Learning for Drug-Induced Liver Injury

QSAR and Predictors of Eye and Skin Effects

An Ex Vivo Perfusion System Emulating In Vivo Conditions in Noncirrhotic and Cirrhotic Human Liver

Contact Info

Product

Resources

About