An Entity-based Visual Analytics System Enhancing Medical Expertise Acquisition: Development and Verification Study (Preprint)

BACKGROUND Acquiring medical expertise from the vast body of medical text is a critical component of medical education. However, most medical knowledge exists in unstructured texts. Data structures vary significantly between medical institutions, resulting in a lack of general-purpose methods for analysis. Furthermore, these data involve patent privacy and are subject to strict confidentiality restrictions, making it difficult to use them directly as learning material or to analyze them with existing web-based tools for knowledge extraction. This creates a significant barrier to medical expertise acquisition for medical learners. OBJECTIVE This study aimed to design, develop, and evaluate MExplore, an interactive visual analytics system to facilitate the acquisition of medical expertise from unstructured medical texts. METHODS We present a workflow for the automatic extraction of medical entities (MEs) from unstructured text, designed for low-cost local deployment to support the analysis of privacy-sensitive medical data. Building on this foundation, we introduce a novel multilevel visual analysis framework that integrates multiple coordinated visualizations to facilitate the acquisition of medical expertise. The framework supports progressive, interactive exploration centered on MEs. To evaluate its effectiveness, we conducted three case studies, a user study, and semi-structured interviews with domain experts. RESULTS MExplore was implemented for medical expertise acquisition. The evaluation demonstrated that the system significantly enhances the process of acquiring medical expertise. Feedback from three case studies, a user study, and expert interviews confirmed that MExplore provides an effective and interactive approach for structurally understanding complex knowledge within medical texts, constructing illness scripts, and strengthening knowledge retention. CONCLUSIONS This paper presents MExplore, an entity-based visual analytics system that effectively supports the acquisition and retention of medical knowledge from unstructured texts. The results prove that MExplore is a valuable tool for medical learners, offering an intuitive and powerful method for in-depth data exploration and analysis.