OBJECTIVES: To identify and evaluate distinct frameworks of potential Generative-Artificial intelligence (Gen-AI) integration in HEOR studies and recommend the most effective approach for maintaining ethical standards, accuracy, trust and efficiency.

METHODS: This study conducts a desk-based conceptual exploration of Gen-AI integration models, identifying and utilizing general Gen-AI concepts and theoretical models, beyond healthcare, to evaluate their application in HEOR and provide recommendations.

RESULTS: Four distinct frameworks of Gen-AI integration potentially useful for HEOR were identified and evaluated — fully human-driven, fully AI-driven, AI-in-loop, and Human Intelligence (HI)-in-loop. The fully human-driven model, which is current status quo, is labor-and-time intensive, subject to human oversights and inefficient for processing large datasets, although it offers high control and accountability. The fully AI-driven model is highly efficient but risks hallucination and lacks trust. The AI-in-loop model provides a balance with some human oversight but risks over-reliance on AI decisions. The HI-in-loop model emerges as the optimal approach, integrating robust human oversight at every stage of the AI process, where the human(s) critically appraises the process and provides continuous feedback at each stage, enhancing AI performance. This builds trust in analyses and ensures adherence to ethical standards required for HEOR studies.

CONCLUSIONS: HI-in-loop model is recommended for HEOR studies to foster collaborative environment as it more effectively combines human expertise with Gen-AI capabilities and ensures decision-making transparency. Emphasizing Gen-AI as a tool to assist, not replace, human capabilities, alongside continuous capacity building, transparent reporting of AI processes, and strict adherence to ethical guidelines, is pivotal. The HI-in-loop approach not only optimizes outcomes but also builds trust among stakeholders, paving the way for more responsible and efficient use of Gen-AI in healthcare research.