From Labels to Immersive (VR) Guidance: Evaluating AI-Assisted Medical Labeling and Immersive IFUs through HCI

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Instructions for Use (IFUs) for transcatheter heart valve (THV) and transcatheter mitral and tricuspid therapy (TMTT) devices are among the most procedurally dense artifacts in medicine – each one must pass through design control, risk analysis, multilingual translation, regulatory submission and post-market surveillance. Two interventions for the same THV/TMTT labelled corpus were evaluated in this paper: (1) AI-assisted authoring and change-impact analysis for labeling content with applicable labeling governing standards, and (2) immersive, procedure-anchored IFU in virtual reality. Representative users assessed the accuracy, completeness, regulatory compliance, traceability, error detection, review efficiency, and usability of the performance of the AI approaches compared to the human approaches. They also completed IFU-related procedures on the device with both the traditional IFU and the immersive guidance experience. Follow up interviews were conducted to see the effect of immersion on comprehension, confidence, attention and perceived authority in labelling information. The study provides a framework for end-to-end HCI, linking the upstream issue of governance around label design to the downstream issue of procedural understanding. The framework does not encourage the use of AI or VR as alternative labels but rather as layers of augmentation that are traceable and under human supervision.