Harnessing Medicinal Chemical Intuition from Collective Intelligence

Area of Science:

• Drug discovery and development
• Computational chemistry
• Medicinal chemistry

Background:

• Decision-making in drug discovery often relies on individual medicinal chemistry intuition.
• Collective intelligence, emerging from group efforts, has shown promise in complex problem-solving.
• The application of collective intelligence to drug discovery, specifically lead optimization, remains underexplored.

Purpose of the Study:

• To explore the application of collective intelligence in drug discovery lead optimization.
• To compare the performance of a collective intelligence agent against an artificial intelligence model.
• To assess the synergy between human and artificial intelligence in drug discovery decision-making.

Main Methods:

• Anonymously surveyed 92 researchers from Sanofi with diverse expertise on ADMET-related questions.
• Developed a collective intelligence agent based on researcher feedback.
• Compared the collective intelligence agent's performance against an artificial intelligence model for various endpoints.

Main Results:

• Collective intelligence significantly improved decision-making for optimizing ADMET (Absorption, Distribution, Metabolism, Excretion, Toxicity) endpoints compared to individual judgments.
• The collective intelligence agent outperformed the artificial intelligence model across most endpoints, with the exception of hERG inhibition.
• A complementary relationship was observed between collective human intelligence and artificial intelligence.

Conclusions:

• Collective intelligence offers a valuable tool for enhancing decision-making in drug discovery lead optimization.
• A synergistic approach combining human and artificial intelligence holds significant potential for improving project decision-making in drug discovery.
• Further research into integrating collective human intelligence with AI can accelerate the drug discovery process.