Artificial intelligence makes surgical ‘black box’ smarter
By: Cynthia Saver, MS, RN
Perioperative services departments have been slow to adopt the black boxes used in the aviation industry, partly because it can take a large team of experts hours to analyze data collected by the box. However, the team behind a surgical black box is using artificial intelligence (AI) to significantly cut analysis time.
“We’ve had a lot of success with automating much of the analysis and processing of the data captured by the OR Black Box®,” says Teodor Grantcharov, MD, PhD, FACS, a professor of surgery at the University of Toronto and holder of the Keenan Chair in Surgery at St Michael’s Hospital, both in Toronto, Ontario, Canada. That data can be used to identify areas of improvement in surgical workflow and technique.
Teodor Grantcharov, MD, PhD, FACS
Teodor Grantcharov, MD, PhD, FACS
The intent behind using the OR Black Box has been to enhance safety in the OR. “It’s giving a voice to the frontline staff who are working in the room every day,” says Catherine Hogan, PhD, RN, program director for perioperative services and infection prevention and control at St Michael’s Hospital.
Currently 12 hospitals in Canada, the United States, and Europe are using the OR Black Box, but AI increases the chances of more widespread use in the ORs of many nurses.
Inside the OR Black Box
The OR Black Box is a tool that supplements ongoing efforts to analyze events in the OR. The main purpose is to identify events leading to intraoperative near misses.
The platform captures and synchronizes intraoperative data from several sources, including the laparoscopic camera (currently, use of the OR Black Box is limited to laparoscopic procedures), panoramic room cameras, audio capturing devices, anesthesia monitors, and other sensors. Over the years, the team has refined the box to enable it to capture large amounts of data that could be used to incorporate AI.
A large library of surgical videos was used to teach the computer, which can now analyze much of a video on its own. For example, because the computer has seen thousands of different types of bleeding, it can now determine whether bleeding is (or isn’t) significant.
Useful analysis depends on collecting data from ORs in a particular location. “Otherwise, we just develop generic one-size-fits-all solutions that don’t always work,” Dr Grantcharov says. “What works in Toronto may not work in New York, Amsterdam, or Copenhagen.”
So, what have OR Black Box data revealed? A 2018 study of 132 elective laparoscopic procedures found that auditory distractions occurred a median of 138 times per case, and at least one cognitive distraction occurred in nearly two-thirds (64%) of cases. The OR door opened about once every 2 minutes. Events, defined as tissue injuries caused by healthcare providers that have potential to cause patient harm, occurred at a median rate of five per hour of operating.
Success stories
St Michael’s Hospital has had OR Black Boxes for the past 4 years. The hospital has six ambulatory ORs and 15 inpatient ORs, two of which have the OR Black Box platform. The plan is to implement the platform in each of the eight new ORs that will soon be added.
Hogan says the OR Black Box has reduced disruption in the OR from people coming in and out. “We gained insight into the number of times the doors were opening, and it prompted the team to think twice about why they are going into a room at that time,” she says. Team members now consider other ways to communicate.
Use of the OR Black Box also has limited the number of people in the room for a case. “As a teaching hospital, we often have people who want to come and watch surgery, but we now look more closely at who is in the room and whether they really need to be there,” Hogan says. “We will ask people to leave the theater if they aren’t necessary.”
The OR Black Box is research based, so there are data to back up recommendations, such as lowering the volume of music played in the OR, Hogan says.
Dr Grantcharov adds that the OR Black Box allows perioperative teams to quantify not only safety threats but also resilience supports (factors that allow systems and teams to be successful despite conditions that can lead to failure) using the Systems Engineering Initiative for Patient Safety Model.
Culture change
“With the OR Black Box, we don’t only focus on things that go wrong,” Dr Grantcharov says. “We also want to capture things that we do exceptionally well.” He notes that clinicians often achieve good patient outcomes even when an adverse event occurs, and there are lessons to be learned from that.
“We want to change the way hospitals usually address quality improvement,” Dr Grantcharov says. Too often, action is taken only after an adverse event, which, he says, creates a natural resistance. “People believe that quality and safety is a bad thing because the only time they talk to quality and safety people is when they make a mistake,” he notes.
A better approach is to use OR Black Box data to reinforce positive behaviors and examine processes that near misses have indicated need to be evaluated. In addition, the entire process should be transparent.
Initial resistance to an OR Black Box is common, but the system doesn’t capture any personal data. “We don’t care who the doctor was, who the nurse was, or who the patient was,” Dr Grantcharov says. “All we want to know is what happened; we’re only interested in the aggregate data.”
Hogan notes that the camera is focused on the room, not the staff. “We were very transparent and showed them what the tape looked like. The OR Black Box is used to improve our practice, our team dynamics, and our communication.” As clinicians see the benefit of the OR Black Box to the team, acceptance increases. “We now have 100% of the general surgeons and 100% of the nurses [at St Michael’s Hospital] on board,” Dr Grantcharov says.
Into the future
“We know that AI will help us put the OR Black Box in every operating room around the world,” says Dr Grantcharov. Challenges to this expansion include computational power. “A lot of these algorithms require very powerful computers that cost hundreds of thousands of dollars,” he says, though he foresees prices dropping to more affordable levels worldwide within a few years.
As with driverless cars, Dr Grantcharov says, it’s simply a matter of time before the OR Black Box will be able to provide real-time suggestions and feedback in the OR, for example, letting the surgeon know when he or she may be about to make a mistake. “We have put ourselves on a timeline of 5 years to develop the first version of real-time feedback,” he says. “We’re confident we can do at least some aspects of it earlier.”
