Deep Learning for Drug-Induced Liver Injury

Area of Science:

• Pharmacology
• Computational Biology
• Toxicology

Background:

• Drug-induced liver injury (DILI) is a primary cause of drug withdrawal.
• Deep learning (DL) offers advanced capabilities for complex data analysis.

Purpose of the Study:

• To develop and validate deep learning models for predicting DILI.
• To identify molecular features associated with DILI.

Main Methods:

• Development of DILI prediction models using deep learning architectures.
• Training and external validation on large drug datasets (475 training, 198 validation drugs).

Main Results:

• The best DL model achieved 86.9% accuracy, 82.5% sensitivity, 92.9% specificity, and an AUC of 0.955.
• Performance surpassed previous DILI prediction models.
• Identification of critical molecular features linked to DILI.

Conclusions:

• DL models show significant potential for improving DILI risk prediction in humans.
• The developed DILI prediction models are publicly accessible for research use.