Deep learning in image-based phenotypic drug discovery.

Area of Science:

• Cellular imaging
• Drug discovery
• Artificial intelligence

Background:

• High-content imaging is crucial for systematic study of chemical compounds' effects on cells.
• Automated screening of cellular images identifies drug-induced phenotypes and potential therapeutic compounds.
• Machine learning (ML), particularly deep learning (DL), has transformed image analysis tasks.

Purpose of the Study:

• To review the applications and adaptations of ML, especially DL, in cell-based phenotypic drug discovery (PDD).
• To highlight how AI accelerates the identification of effective drugs and their modes of action.

Main Methods:

• Review of current literature on ML and DL in PDD.
• Focus on image analysis techniques for identifying cellular phenotypes.
• Adaptation of AI methods for large-scale image screening in drug discovery.

Main Results:

• ML and DL methods show significant promise in enhancing PDD.
• AI accelerates the identification of 'hit' compounds with therapeutic potential.
• These methods aid in understanding drug-induced cellular changes and mechanisms of action.

Conclusions:

• AI, especially DL, is a disruptive technology in phenotypic drug discovery.
• ML-driven image analysis offers powerful tools for identifying novel therapeutics.
• The integration of AI is poised to significantly impact future drug discovery pipelines.