Development of a Video-based Personal Protective Equipment Monitoring System

PROJECT SUMMARY
During the COVID-19 pandemic, healthcare workers (HCWs) have had a more than 11-fold higher infection
risk than the general population. Several risk factors for COVID-19 infection among HCWs have been
identified, including the lack of personal protective equipment (PPE) and inadequate PPE use. Among these
factors, the inadequate use of PPE has been associated with a one-third higher risk of infection. Given the high
incidence of infection, there is a critical need to address the challenges of monitoring and promoting adherence
with appropriate PPE use among HCWs. The long-term goal of this research is to reduce workplace-acquired
infections in HCWs by improving adherence to appropriate PPE use in settings at high risk of transmission.
The overall objectives of this proposal are to design, implement, and test a system (Computer-Aided PPE
Nonadherence Monitoring and Detection—CAPPED) that (1) tracks the team’s PPE adherence using computer
vision and (2) highlights episodes of potential PPE nonadherence on a video-monitoring system. Our central
hypothesis is that continuous monitoring of PPE use by multiple HCWs is a complex, cognitively demanding,
and error-prone task unaddressed by current methods for monitoring PPE adherence. The rationale for this
proposal is that enhanced recognition of PPE nonadherence is a requirement for reducing transmissible
infections in HCWs. Guided by preliminary data, the central hypothesis will be tested by pursuing two specific
aims: (1) design and implement a computer vision system (CAPPED) for recognizing PPE nonadherence in a
dynamic, team-based setting, and (2) compare human performance during simulated resuscitations using
direct observation, basic video surveillance, and computer-aided monitoring (CAPPED system). For the first
Aim, machine learning approaches will be applied to recognize the type of nonadherent PPE (headwear,
eyewear, mask, gown, gloves) and the category of nonadherence (absent or inadequate). Under the second
Aim, a customizable visual interface will be designed and evaluated for monitoring and spotlighting PPE
nonadherence with a human-in-the-loop. The proposed research is innovative because it addresses the
challenges of simultaneously identifying nonadherence with several types of PPE used by multiple individuals
in a dynamic setting. This proposed research is significant because it is expected to reduce infection
transmission to HCWs by tracking and eventually alerting them to nonadherent PPE use. The results of this
research are expected to positively impact the workplace safety of HCWs by addressing the limitations of
current approaches to PPE monitoring.

Grant Number: 5R01EB032819-03
NIH Institute/Center: NIH
Principal Investigator: RANDALL BURD