Thesis
Designing explainable user interfaces for video-based PPE compliance monitoring in time critical medical events
Master of Science (M.S.), Drexel University
Jun 2026
DOI:
https://doi.org/10.17918/00011443
Abstract
Personal protective equipment (PPE) compliance monitoring is critical for reducing healthcare-associated infections (HAI) in high-risk clinical settings such as trauma resuscitation. Computer vision (CV) systems offer an automated, scalable approach to monitoring PPE compliance; however, existing systems prioritize model accuracy over the design of explainable user interfaces (XUI) that make CV outputs interpretable for clinical users. This thesis addresses this gap by applying the principles of human-centered design (HCD) and XUI to the iterative design and evaluation of a video-based XUI for real-time monitoring of PPE compliance in trauma resuscitation. Through a heuristic evaluation and two rounds of structured remote usability testing with 10 clinical domain experts -- including nurse educators, ICU nurses, and nursing students -- four versions of the XUI (V1-V4) were designed and evaluated. The XUI communicates CV model outputs through icon-based visual explanations, progressive disclosure of information, and cross-panel referencing to support rapid detection of PPE noncompliance. Findings across both rounds of usability testing show that familiar, domain-aligned iconography with red and yellow color-coding supported rapid visual recognition of PPE noncompliance. Moving away from CV outputs likebounding boxes helped reduce visual clutter, while the hover interaction modality and auto-scroll enabled rapid PPE noncompliance detections. V3.3 of the XUI achieved an overall ease of use mean of 4.6 out of 5 (1 = very difficult, 5 = very easy), and the final XUI (V4) incorporated hover interactions and auto-scroll based on participant feedback. The primary contribution of this thesis is identifying visual clutter in video-based clinical XUIs as a fundamental challenge rooted in information density rather than visual design alone, requiring interaction-level solutions. This thesis proposes four design principles for video-based clinical XUIs: progressive disclosure through interaction modalities, workflow sensitivity, root cause analysis for usability issues, and multiple rounds of user evaluation. These principles extend existing XUI design frameworks and offer actionable guidance for designers building explainable AI interfaces for high-stakes, video-based, information-dense clinical environments.
Metrics
1 Record Views
Details
- Title
- Designing explainable user interfaces for video-based PPE compliance monitoring in time critical medical events
- Creators
- Vidhi Shah
- Contributors
- Aleksandra Sarcevic (Advisor)
- Awarding Institution
- Drexel University
- Degree Awarded
- Master of Science (M.S.)
- Publisher
- Drexel University
- Number of pages
- xvi, 70 pages
- Resource Type
- Thesis
- Language
- English
- Academic Unit
- Digital Media; Drexel University; Antoinette Westphal College of Media Arts and Design
- Other Identifier
- 991022189365304721