Physicians and computer scientists are installing depth sensors and various monitors in hospital hallways, near patients' bedsides and in operating rooms to monitor staff 24/7 and improve hand hygiene compliance, The Wall Street Journal reports.
Seven things to know:
1. The sensors produce video images resembling blurry silhouettes to protect patient privacy, but can train computer algorithms to identify movements, such as a staff member stopping to use a hand sanitizer dispenser.
The technology, called computer vision, would allow hospitals to monitor workers 24/7 across an entire hospital ward. The researchers aim to use the data from the algorithms to influence staff behaviors and rethink how they administer care.
2. Even the most attentive physicians and nurses may be unaware when they skip steps in care, particularly when that can involve hundreds of small tasks in a day, said Arnold Milstein, MD, professor of medicine and director of the Clinical Excellence Research Center at Stanford (Calif.) University.
To address this issue, Dr. Milstein and his colleagues created a series of pilot studies to determine whether computer vision could tell if staff used hand sanitizer dispensers before entering and leaving patient rooms.
3. The researchers installed sensors above the hand-sanitizer dispensers in a hallway of an acute-care ward at Palo Alto, Calif.-based Lucile Packard Children's Hospital. They collected thousands of images and annotated about 80 percent of them, noting whether someone did or did not sanitize their hands when entering or leaving a patient's room.
4. The researchers fed the annotated images into an algorithm to teach it to distinguish whether someone was washing hands. They used the remaining images to see whether the algorithm could identify handwashing without using the annotations.
5. The algorithm was then applied to images gathered at an intensive care unit of Salt Lake City-based Intermountain Healthcare. Although the hospital's hallway was different than the other hospital's, the algorithm identified hand sanitization roughly 85 percent of the time. When the algorithm was further trained on images Intermountain Healthcare captured, its accuracy jumped to 98 percent, Serena Yeung, a PhD student in the Stanford Artificial Intelligence Laboratory, told WSJ.
6. Now, the researchers are determining how the data can be used to promote vigilance in handwashing. One idea includes having an alert on the sanitizer dispensers to remind staff to wash their hands before entering patient rooms. Another idea is to design a digital dashboard to track compliance across the unit..
"People would know how they are performing as a group and hopefully be motivated to improve," Ms. Yeung said. She suggested training programs or rewards for units with the best compliance records.
7. Additionally, researchers aim to use computer vision to study handwashing protocols inside patient rooms to determine whether current practices, such as washing hands after touching a patient, are effectively fighting harmful bacteria or if different steps should be considered.