A function-based framework for AI-amplified, data-driven healthcare simulation research

Abstract Healthcare simulation research increasingly generates rich, multimodal datasets, yet substantial portion of these data remain underutilized due to analytic, technical, and resource constraints. As artificial intelligence becomes more accessible, there is a risk that adoption in simulation research is driven by novelty rather than by clearly articulated analytic needs. This article proposes a function-based framework that conceptualizes artificial intelligence as a research amplifier for data-driven simulation-based research. Rather than organizing approaches by algorithm family or tool type, the framework groups AI use according to its function in the research workflow across four domains: (1) Data Processing and Integration: transforming raw traces into synchronized, research-ready datasets; (2) Comparative and Predictive Analytics: benchmarking performance and modeling learning trajectories or outcomes; (3) Behavioral and Interaction Analysis: decoding individual and team behaviors from observational traces; and (4) Automation and Acceleration: streamlining research activities such as transcription, coding support, and structured summarization through automated or semi-automated workflows. Five practical steps illustrate how the framework can be embedded in research planning: clarifying the analytic task, mapping it to functional domains, specifying data representations and integration needs, designing the human–AI division of labor, and planning evaluation and reporting. Ethical, trust, and accountability considerations are discussed across functional domains, emphasizing that distinct risks and safeguards emerge across different AI-enabled research activities. By centering function and purpose, this framework aims to help simulation researchers make explicit, intentional decisions about when and how AI meaningfully extends analytic capacity.