Researchers warn of bias as Montreal health network to use AI to reduce wait times

Hey there, time traveller!
This article was published 29/12/2022 (680 days ago), so information in it may no longer be current.

MONTREAL – Artificial intelligence researchers say a Montreal hospital’s plan to reduce emergency room wait times with an AI algorithm is an appropriate use of the technology — if it’s done carefully.

The Centre hospitalier de l’Université de Montréal, one of the city’s two main hospital networks, is testing an artificial intelligence algorithm intended to help administrators plan emergency room staffing and accelerate the admission of patients.

The health centre says the AI system will use data from the past 20 years to predict when its emergency rooms will be particularly busy, allowing the network to increase staffing levels on certain days and schedule elective surgeries when fewer patients are expected.

Abhishek Gupta, founder of the Montreal AI Ethics Institute, says algorithms can be useful to help reduce wait times, but he warns that the hospital will have to be careful to avoid perpetuating biases.

“For example, if historical patient visits are going to be used as the data source, an analysis to understand if there are any pre-existing biases will help to avoid baking them into the system,” he wrote in an email Thursday. It’s important, he added, that patients be told how their data will be used and stored.

Bias is also a concern for Fenwick McKelvey, a communications studies professor at Concordia University who studies digital policy.

“We know that there’s systemic racism in the Quebec medicare system,” he said in an interview, adding that the 2020 death of Joyce Echaquan drew attention to discrimination in the province’s health network.

Echaquan, an Indigenous woman, filmed herself on Facebook Live as a nurse and an orderly were heard making derogatory comments toward her at a hospital in Joliette, Que., northeast of Montreal, shortly before her death. A coroner concluded Echaquan did not receive the care she needed because prejudice contributed to a faulty diagnosis.

Dr. Élyse Berger Pelletier, an emergency room physician working on the AI project, said that with Quebec patients waiting an average of 18 hours between the time they’re admitted by a doctor and when they’re given a bed on a ward, there’s a need to work more efficiently.

“I’m an emergency physician working in the emergency room full time; I see how much it deteriorates, how much we want to give quality care and that we’re not always able to do it the way we want,” she said in an interview. “So, to be able to work with tools that will make our life easier, for me, it’s a solution that is urgent.”

Another element of the system, which is being developed by an in-house research team, will consider factors like a patient’s age and symptoms to determine how likely they are to be admitted, allowing doctors to request a bed for a patient before all the usual tests are completed, Berger Pelletier said.

“This is really where the value for the patient is, because we don’t want them to wait and we know that when you stay on a stretcher in the ER, particularly for elderly people, it’s not good for them; we know that they have an increase in mortality and morbidity,” Berger Pelletier said.

Berger Pelletier said she expects the system to officially launch within the next year and that some elements could be deployed in six months.

As well, she said she takes the risk of bias seriously. Considering that the AI tool will be used to manage staffing levels and assign beds, there’s less chance of harm than if it was being used to determine what type of care patients receive, she said.

“It’s not treating patients; it’s about managing a hospital.”

Berger Pelletier said the algorithm will be regularly monitored to ensure it’s working, something that Gupta said is necessary for AI systems.

But while the potential use of AI in health care tends to draw attention, McKelvey said, he worries technology is only a Band-Aid solution to deeper problems in Canada’s health-care system.

“I certainly welcome innovation in delivery, but that doesn’t seem to fix the more deeper structural issues that seem to be at work in the medicare system across Canada.”

But Berger Pelletier said she thinks technologies like artificial intelligence will become increasingly important as Quebec’s population ages. In particular, she sees the opportunity for technology to help free health-care workers from clerical duties so they can focus on patient care.

“If we want to treat everyone adequately and with quality, the only way is to have technology to help the humans, so that the human remains in contact with the patient,” she said.

This report by The Canadian Press was first published Dec. 30, 2022.