AI-Driven Multiagent System for Guiding First-Line Immunotherapy for NSCLC

Objective:

To improve predictive tools for guiding first-line immunotherapy decisions in patients with previously untreated non-small cell lung cancer (NSCLC), addressing the limitations of current biomarker-based approaches.

Key Findings:

• The system produced correct treatment recommendations 72% of the time.
• Recommendations were deemed helpful 72% and complete 91% of the time.
• Retrieved information was meaningful in 98% of queries.
• 6% of statements were found to be harmful.

Interpretation:

The multi-agent system shows promise in supporting clinical decision-making for immunotherapy in NSCLC, although further validation and enhancement of trustworthiness are needed.

Limitations:

• The study was small and conducted at a single institution.
• Tool usage was correct in only 56% of cases, indicating potential areas for improvement in system reliability.

Conclusion:

The study serves as a proof-of-concept for agentic AI in clinical decision support, highlighting the need for standardized evaluation frameworks and further studies to assess the impact of AI systems as they become more autonomous.