Objective:

To develop and evaluate an LLM-based chatbot framework for colorectal cancer (CRC) screening.

Approach:

• Assessment of LLM Performance: Evaluate the baseline performance of contemporary LLMs in CRC screening communication.
• Optimization of Chatbot: Optimize chatbot performance through structured prompt engineering and integration.

Key Findings:

• CRC screening uptake is suboptimal due to low awareness and psychological barriers.
• Evidence supporting the use of LLMs in CRC screening is limited.
• Most existing studies focus on answering CRC-related questions rather than structured screening workflows.

Interpretation:

The study emphasizes the importance of aligning LLM outputs with established screening guidelines to ensure clinical applicability.

Limitations:

• Insufficient validation of LLM performance in structured screening workflows.
• Concerns regarding the quality and reliability of AI-generated medical information.

Conclusion:

The study aims to address the gaps in LLM application for CRC screening by developing a structured chatbot framework.

Sources:

• Global cancer statistics 2020
• Screening for colorectal cancer: US Preventive Services Task Force recommendation statement
• Performance of large language models in addressing patient queries on colorectal cancer screening