Large Language Models in Clinical Trial Recruitment: Sociotechnical and Economic Framework Development Study

Area of Science:

• Clinical AI
• Health Informatics
• Sociotechnical Systems

Background:

• Large language models (LLMs) demonstrate potential for patient-trial matching.
• Current evaluations often lack real-world workflow integration.
• Organizational factors like privacy and oversight significantly impact LLM deployment outcomes.

Purpose of the Study:

• Develop a theory-grounded framework for integrating LLM-enabled clinical AI into recruitment.
• Analyze the impact of this integration on operational, governance, and economic aspects.
• Provide a structured approach for evaluating LLM implementation in clinical trial recruitment.

Main Methods:

• Structured conceptual analysis and targeted evidence synthesis.
• Integration of sociotechnical systems theory and transaction cost economics.
• Development of the LLM-Embedded Clinical Recruitment Architecture (LECRA) framework.

Main Results:

• LECRA models recruitment as a closed-loop sociotechnical and economic system.
• Identifies key moderators of performance: privacy, hallucination risk, bias, patient trust, oversight, and regulatory costs.
• Reframes recruitment performance as multidimensional and proposes an empirical testing roadmap.

Conclusions:

• LECRA provides a deployment-sensitive analysis of LLM-enabled recruitment value.
• Highlights potential benefits versus coordination, compliance, and trust-related frictions.
• Aims to support future empirical studies and realistic implementation decisions.