Currently submitted to: JMIR Nursing
Date Submitted: Mar 24, 2026
Open Peer Review Period: Apr 8, 2026 - Jun 3, 2026
(currently open for review)
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
A Computer Vision–Based Monitoring Prototype for Unplanned Extubation Prevention: Design, Development, and Preliminary Evaluation
ABSTRACT
Background:
Unplanned extubation (UEX) in the intensive care unit (ICU) is a serious adverse event that threatens patient safety. Current prevention strategies rely primarily on nurse surveillance and physical restraints—both of which have inherent limitations, including the inability to provide continuous monitoring and potential conflicts with patient-centered care principles. Computer vision technology may offer a noncontact monitoring approach to support nursing practice.
Objective:
This study aimed to develop and preliminarily evaluate a computer vision–based monitoring prototype designed to assist ICU nurses in preventing UEX by detecting patient hand proximity to preset high-risk zones, with particular emphasis on a guard mode that detects protective gear removal.
Methods:
Following a design science research approach, we built a monitoring system using the open-source MediaPipe library. The system comprised four components: (1) patient tracking with a three-level detection mechanism, (2) hand detection and risk assessment with tiered alerting, (3) dynamic risk zone management, and (4) a PyQt5-based user interface. A guard mode was implemented to specifically monitor for protective gear removal—a critical feature supporting stepwise physical restraint reduction. Laboratory simulations were conducted under various occlusion scenarios. Five ICU nurses participated in usability assessments using a 5-point Likert scale. Performance metrics included tracking accuracy, hand detection coverage, risk judgment accuracy, false alarm rate, and alert response time.
Results:
Across 120 simulated test scenarios, target localization accuracy reached 92.3%, and dynamic risk zone adaptation accuracy was 90.0%. Hand recognition coverage was 95.8% across 800 frames. Two-level risk judgment accuracy was 93.3%, with the false alarm rate reduced to 2.8%. The average alert response time was 0.4 seconds. In guard mode, the system successfully distinguished between gloved hands (protective gear) and bare hands, triggering dedicated alerts upon detection of gear removal. Nurses rated the system’s usability at 4.2 out of 5. Figures are provided to illustrate the system interface and the visual differences in hand monitoring between normal and guard modes.
Conclusions:
This study describes the development and preliminary evaluation of a computer vision–based system for UEX prevention. Our findings suggest the system is technically feasible for supporting nursing surveillance. The guard mode, in particular, offers a novel approach to supporting gradual restraint reduction—a key priority in patient-centered care. Further clinical validation is needed to assess its effectiveness in practice settings. Clinical Trial: None
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.