In Silico Hazard Assessment of Ototoxicants Through Machine Learning and Computational Systems Biology

Area of Science:

• Environmental Health
• Toxicology
• Computational Biology

Background:

• Chemicals and medications can cause hearing loss (ototoxicity) across all life stages.
• Environmental exposures raise concerns regarding potential ototoxic effects.

Purpose of the Study:

• To apply computational modeling for hazard identification and chemical prioritization concerning ototoxicity.
• To screen environmental chemicals for potential ototoxic properties using in silico models.

Main Methods:

• Evaluated in silico drug-induced ototoxicity models on 76 environmental chemicals.
• Utilized fingerprinting, similarity assessment, and machine learning classification.
• Constructed an ototoxicity pathway using a systems biology framework for PCB 177 and mitotane.

Main Results:

• A consensus computational model achieved high accuracy (0.95 test; 0.90 validation).
• Identified 18 of 76 chemicals as potential ototoxicants.
• Predicted overlapping mechanisms, including thyroid-stimulating hormone receptor signaling disruption by mitotane and PCB 177.

Conclusions:

• Integrated computational modeling effectively supports virtual screening for ototoxicity risk assessment.
• Findings suggest thyroid-mediated pathways may contribute to chemical-induced auditory harm.
• This approach aids in prioritizing chemicals for further ototoxicity evaluation.