Explainable Artificial Intelligence: A Perspective on Drug Discovery

Area of Science:

• Computational chemistry
• Pharmacology
• Artificial intelligence

Background:

• Artificial intelligence (AI) and deep learning (DL) accelerate drug discovery but suffer from a "black-box" problem, hindering pharmaceutical researcher adoption.
• Explainable artificial intelligence (XAI) offers a solution by increasing transparency and trust in AI models.

Purpose of the Study:

• To systematically review XAI principles, methodologies, and tools for drug discovery.
• To explore XAI applications in accelerating various stages of the drug discovery pipeline.
• To examine how XAI addresses AI model opacity and its challenges.

Main Methods:

• Systematic investigation of XAI principles and methodologies.
• Review of XAI tools, models, and frameworks for drug discovery.
• In-depth discussion of XAI applications in healthcare and drug development.

Main Results:

• XAI enhances transparency, trust, and reliability in AI-driven drug discovery.
• XAI applications span molecular modeling, target identification, ADME prediction, and clinical trial design.
• XAI effectively bridges the gap between computational predictions and pharmaceutical applications.

Conclusions:

• XAI is crucial for overcoming the "black-box" nature of AI in drug discovery.
• Further research is needed to address challenges in deploying XAI methodologies.
• Staying updated on XAI technologies is vital for improving drug discovery efficiency and clinical impact.