In a busy ward at St. Michael’s Hospital in downtown Toronto, one of Shirley Bell’s patients was suffering from a cat bite and a fever but otherwise appeared healthy — until an alert from an AI-powered early warning system showed he was sicker than he appeared.
While the nursing team would typically check blood test results around noon, the technology would announce the results several hours in advance. That warning indicated that the patient’s white blood cell count was “really, really high,” recalls Bell, a clinical nurse educator for the hospital’s general practice program.
The cause turned out to be cellulitis, a bacterial skin infection that, if left untreated, can lead to significant tissue damage, amputations and even death. Bell said the patient was given antibiotics quickly to avoid those worst-case scenarios, thanks in large part to the team’s in-house artificial intelligence technology, called Chartwatch.
“There are many, many other scenarios where patients’ conditions are reported earlier, the nurse is alerted earlier, and interventions are put in place earlier,” she said. “This doesn’t replace the nurse at the bedside; it actually enhances your nursing care.”
A year and a half study on Chartwatch, published Monday A study published in the Canadian Medical Association Journal found that using the AI system led to a dramatic 26% drop in unexpected deaths among hospitalized patients.
“We are pleased to see that we are saving lives,” said co-author Dr. Muhammad Mamdani, vice-president of data science and advanced analytics at Unity Health Toronto and director of the Centre for Research and Education in AI in Medicine at the University of Toronto’s Temerty Faculty of Medicine.
“A promising sign”
The research team examined more than 13,000 admissions to St. Michael’s general internal medicine ward — an 84-bed unit that cares for some of the hospital’s most complex patients — to compare the tool’s impact among this patient population to thousands of admissions to other subspecialty units.
“Over the same time period, in other units in our hospital that were not using Chartwatch, we did not see a change in these unexpected deaths,” said lead author Dr. Amol Verma, a clinician-scientist at St. Michael’s, one of three sites in the Unity Health Toronto hospital network, and the Temerty Professor of AI Research and Education in Medicine at the University of Toronto.
“It was a promising sign.”
Unity Health’s AI team began developing Chartwatch in 2017, based on suggestions from staff that predicting death or serious illness could be key areas where machine learning could make a positive difference.
The technology underwent several years of development and rigorous testing before being rolled out in October 2020, Verma said.
“Chartwatch measures about 100 pieces of data from a patient’s medical record that are routinely collected as part of healthcare delivery,” he explained. “So a patient’s vital signs, their heart rate, their blood pressure… all the lab results that are done every day.”
Working in the background alongside clinical teams, the tool monitors any changes in a person’s medical record “and makes a dynamic prediction every hour about how likely that patient’s condition is to deteriorate in the future,” Verma told CBC News.
This could mean someone is getting sicker, needs intensive care, or even is on the verge of death, giving doctors and nurses a chance to intervene.
In some cases, these interventions involve stepping up a person’s level of treatment to save their life, or providing early palliative care in situations where patients cannot be saved.
In both cases, the researchers said, Chartwatch appears to complement clinicians’ judgment and lead to better outcomes for frail patients, helping to prevent more sudden and potentially preventable deaths.
AI booming in healthcare
Beyond its uses in medicine, artificial intelligence has generated a lot of buzz – and backlash – in recent years.
From the controversy surrounding the use of machine learning software to produce academic essays, to concerns about AI’s ability to create realistic audio and video content that mimics real celebrities, politicians, or ordinary citizens, there are many reasons to be cautious about this emerging technology.
Verma himself said he’s long been wary. But in health care, he said, these tools have immense potential to address the staffing shortages plaguing Canada’s health care system by complementing traditional bedside care.
Often considered the future of health care, artificial intelligence is already finding its way into Canadian hospitals. But AI is far from perfect, and some are concerned about the costs it could entail.
These efforts are still in their early stages. Various research teams, including private companies, are exploring ways to use AI to detect cancer earlier. Some studies suggest it has the potential to report hypertension simply by listening to someone’s voice; others show that it could scan brain patterns to detect signs of a concussion.
Chartwatch is remarkable, Verma said, because of its success in keeping real patients alive.
“Very few AI technologies have been implemented in clinical settings until now. To our knowledge, this is one of the first in Canada that has actually been implemented to help us care for patients on a daily basis in our hospital,” he said.
A ‘real-world’ look at AI’s impact on healthcare
The study at St. Michael’s Hospital has limitations, however. It was conducted during the COVID-19 pandemic, at a time when the health-care system was facing a range of unusual challenges. The team also acknowledged that the urban hospital’s patient population is also unique, given the high level of complex patients, including those facing homelessness, addiction and related health issues.
“Our study was not a randomized controlled trial conducted across multiple hospitals. It was conducted within a single organization, within a single unit,” Verma said. “Before we can say this tool can be used everywhere, I think we need to do research on its use in multiple settings.”
Dr. John-Jose Nunez, a psychiatrist and researcher at the University of British Columbia who was not involved in the study, agreed that the research needs to be replicated elsewhere to get a better idea of how effective Chartwatch is in other settings. He also added that patient privacy needs to be considered when using new AI technologies.
Still, he praised the study team for providing a “concrete” example of how machine learning can improve patient care.
“I really think AI tools are becoming an additional member of the clinical care team,” he said.
The Unity Health team hopes its technology will be deployed more widely in the future, within its own Toronto-based hospital network and beyond.
Much of this work is done through GEMINICanada’s largest hospital data sharing network for research and analytics, said Mamdani, Unity Health’s vice president of data science.
More than 30 Ontario hospitals are working together, he said, providing opportunities to test Chartwatch and other AI tools in various clinical and hospital settings.
“This now lays the foundation to be able to deploy these things well beyond our four walls,” Mamdani said.
