User-Centered Design of a Workflow for Multidisciplinary Medical Dataset Creation

Zsófia Gergely

Supervisor(s): Erika Váczlavová

Slovak Technical University

Abstract: The increasing availability of multimodal medical data, including whole slide images (WSIs), has intensified collaboration between clinical and information technology (IT) domains. However, medical domain experts (DEs) and IT specialists approach dataset creation with fundamentally different goals. Their divergent mental models complicate decision-making throughout the dataset construction process. This paper presents a human-centered workflow, grounded in user experience (UX) principles, for multimodal medical image dataset creation in digital pathology. The workflow is structured in two interconnected flows: a human-centered workflow for mental model analysis focused on identifying and mapping the mental models of medical DEs and IT specialists, and a technical workflow for dataset creation covering digitization, anonymization, annotation, and validation, with standardization integrated at specific stages. Grounded in structured user research involving mental model analysis, workflow mapping, and persona synthesis, the human-centered workflow uncovers the differences in expectations between the two domains. These insights guide the technical workflow, where each step is informed by human-centered evaluation to determine solutions that accommodate both clinical and computational needs. By systematically integrating the perspectives of different domain experts into each step, the approach supports the creation of medical datasets that are both clinically meaningful and technically usable, enabling effective collaboration and reuse across domains.
Keywords: Human-Computer Interaction
Full text:
Year: 2026