Pilot study discovers computer vision technology

In early 2020, before a COVID-19 vaccine and effective treatments were widely available, universal mask-wearing was a central strategy to prevent transmission of COVID-19. However, hospitals and other facilities where masks are mandatory have faced challenges. Reminding patients, visitors, and employees to wear masks had to be done manually, which was time-consuming and labor-intensive. Researchers at the Massachusetts Institute of Technology (MIT) and the Massachusetts Institute of Technology (MIT) set out to test a tool that uses computer vision algorithms to automate mask-wearing monitoring and reminders. The team conducted a pilot study with volunteered hospital employees and found the technology to work effectively, with most participants reporting positive experiences operating the system at hospital entrances. I discovered that The results of this study are BMJ Open.

“Changing behaviors such as mask wearing requires a lot of effort, even among health professionals,” said lead author Peter Chai, MD, MMS, Emergency Medicine. “Our research suggests that such a computer visualization system could prove useful during the next respiratory virus pandemic, in which masks could be used in a hospital setting. It is an essential strategy for controlling the spread of infection.”

“We are aware of the challenges in ensuring proper mask use and the potential barriers associated with HR-based notification of mask misuse by colleagues. I will discuss the proposal and my colleagues’ evaluation of the platform’s initial acceptability,” said the senior author. He is C. Giovanni Traverso, MB, BChir, PhD, from BWH’s School of Medicine and MIT’s School of Mechanical Engineering.

For this study, the team used a computer vision program developed using low-resolution closed-circuit television still frames to detect mask wearing. Between April 26, 2020 and April 30, 2020, researchers will participate in an observational study testing computer vision models of employees trying to enter one of the hospital’s main entrances. I invited you to The team enrolled his 111 participants who operated the system and surveyed them about their experiences.

The computer visualization system accurately detected mask adhesion 100% of the time. Most participants (87%) reported having a positive experience operating the system at the hospital.

The pilot was limited to one hospital employee and may not be generalizable to other settings. Additionally, behaviors and attitudes toward masking have changed over the course of the pandemic and may vary across the United States. Further research is needed to identify the barriers.

“Our data suggest that individuals in hospital settings are embracing the use of computer visualization systems to help detect effective mask wearing and provide reminders. As a way to keep ourselves safe on the front lines of a medical emergency,” Chai said. “Continued development of detection systems may provide useful tools in the context of the COVID-19 pandemic or as a preparedness to prevent future spread of airborne pathogens.”

Disclosure: Traverso has a financial interest in Teal Bio, a biotechnology company that manufactures transparent, reusable ventilators. Adam Landman is a member of the Abbot Medical Device Cybersecurity Council. For full details of all of Traverso’s commercial and non-commercial relationships, please see the following links: The other authors have no competing interests to report.

Funding: Funding for this study was provided by the National Institutes of Health (K23DA044874, R44DA051106) Hans and Mavis Loper Psychosocial Foundation, Carl Van Tassel (1925) MIT, Department of Mechanical Engineering, MIT and Department of Gastroenterology, Brigham and Women’s Hospital.

Papers cited: Chai PR and others. “Acceptance of Computer Vision Facilitated Protocols for Measuring Face Mask Adherence: A Single-Site Observational Cohort Study Among Hospital Staff” BMJ Open DOI: 10.1136/bmjopen-2022-062707

Disclaimer: AAAS and EurekAlert! EurekAlert! is not responsible for the accuracy of news releases posted. Use of information by contributors or via the EurekAlert system.