Could artificial intelligence help Quebec’s cash-strapped health-care system free up resources to better serve patients even amid the enduring workforce shortage?
A new software platform being used to optimize cancer treatments at Montreal’s CHUM superhospital and other facilities in Quebec is giving officials hope that durable efficiency improvements can be made across the health network.
Dubbed GrayOS, the software was developed by Montreal-based startup Gray Oncology Solutions over a three-year period in partnership with the CHUM. Authorities say the new system has helped improve access, shorten waiting lists and allow employees such as nurses to spend more time caring for patients — all the while improving employee morale by removing some tedious tasks from their workload.
“People think that booking an appointment in a hospital is simple, but actually it’s ultra-complicated,” Kathy Malas, head of innovation and AI at the CHUM, told the Montreal Gazette. “There are upwards of 25 variables to consider so that the people who are responsible for these appointments, the health-care technicians or administrative agents, can book one. These constraints include the availability of the patient and the doctor, the type of technology, the length of the intervention and the room involved.”
While health-care staffers “are very smart, there is a limit to the amount of data that they can process to optimize the appointment,” Malas added. “It’s like your Outlook agenda. When you do it manually, it’s very time-consuming, and you could end up with suboptimal programming, meaning holes in the schedule.”
GrayOS is directly integrated into the CHUM’s existing software. That enables automated and optimized treatment planning for cancer patients, the hospital says.
An algorithm built into the software allows health professionals to determine patient treatment times before treatment begins. An online patient scheduling tool constantly evaluates patient flow and finds the best start date for each new arrival.
Results so far have been positive, Malas said. They include a five per cent increase in efficiency at an infusion clinic, which translates into 11 hours of additional capacity daily to treat patients without hiring new employees.
GrayOS’s optimization tool let CHUM staff book more than 5,000 additional appointments during COVID-19’s second wave, according to the hospital. The software also cut the administrative burden of employees by 80 per cent, resulting in higher worker satisfaction, fewer burnouts and less employee turnover, the CHUM says.
“The time savings have been enormous,” Malas said. “When you can reinvest this much time in caring for patients, it’s a big deal. If you are a clinician, what you want is to spend as much time as possible in direct patient care. Booking appointments is not what drives you.”
Malas will be one of the speakers Wednesday at All In, a two-day conference on artificial intelligence that’s poised to attract more than 1,400 industry experts, entrepreneurs, government officials and investors from 17 countries to Montreal. Her panel will look at how AI is changing the health-care sector.
Leading scientists such as deep-learning guru Yoshua Bengio — who will also be speaking at the conference — have repeatedly warned about the dangers of AI and have urged governments to legislate its use.
The adoption of GrayOS follows a decision in 2018 by the CHUM to use AI ethically.
“AI was already in our cellphones, and we wanted to prepare our organization to integrate it in an ethical and responsible way,” Malas said. “The goal was to develop and adopt technologies that create tangible benefits for patients, our teams and the organization. When the risks were too great or the benefits insufficient, we stopped everything.”
Data breaches represent one of the biggest risks surrounding the use of AI in health care. Hospitals collect substantial amounts of confidential patient data, which makes them attractive targets for cybercriminals.
In the case of GrayOS, risks of a data breach or a loss of confidentiality “were low, so we proceeded quickly,” Malas said. “The data stayed on our local servers and employees were granted access according to the highest standards. The benefits far outweighed the risks.”
About 35 people now work for CITADEL, the hospital’s in-house centre for medical data integration and analysis. Some 106 AI-driven projects — some of which were dropped due to ethical or legal concerns — have been launched at the CHUM since 2018, Malas said.
Appointment optimization tools, meanwhile, could soon become a reality in other CHUM departments.
“GrayOS is very popular at CHUM now,” Malas said. “People want Gray in departments like radiology and angiography. The notion of optimizing appointments in a hospital is pertinent for several sectors. Our challenge is to bring it up to scale beyond oncology. The variables are different, so we’re going to need to retrain the algorithms. We can’t do cut-and-paste because the constraints are different.”
AI guru says Canada’s regulation is too slow, warns of ‘existential’ threats
Powerful new chatbots highlight need for legislation on AI, Montreal conference told
Five per cent of Quebec businesses use AI: report