Tackling the complexity of cancer with generative models

Area of Science:

• Oncology
• Artificial Intelligence
• Computational Biology

Background:

• The Hallmarks of Cancer framework provides a reductionist view of cancer biology.
• This framework unifies observations and offers mechanistic insights but doesn't fully capture cross-scale interactions.
• A need exists for tools that address cancer's complex, multimodal, and multiscale nature.

Purpose of the Study:

• To propose generative artificial intelligence (AI) models as a key technology for understanding and intervening in cancer.
• To highlight the potential of generative AI to capture cancer's complexity.
• To envision a synergistic relationship between generative AI and the Hallmarks of Cancer framework.

Main Methods:

• Leveraging recent advances in artificial intelligence, specifically generative models.
• Utilizing the pattern recognition, unstructured input processing, and multimodal synthesis capabilities of generative AI.
• Integrating generative AI with the established Hallmarks of Cancer framework.

Main Results:

• Generative models can recognize complex patterns in biological data.
• These models can process unstructured and synthesize multimodal inputs relevant to cancer.
• A synergistic cycle is envisioned where generative AI drives discovery and the Hallmarks guide measurement development.

Conclusions:

• Generative AI models are poised to revolutionize cancer diagnosis, understanding, and intervention.
• These models offer a complementary approach to the Hallmarks of Cancer, addressing its limitations in capturing complexity.
• The integration of generative AI promises a new era in biological discovery and clinical care for cancer.