“All innovations must make sense and bring tangible benefits to patients, healthcare teams and, ultimately, to society as a whole. » Kathy Malas, the young director of the health innovation and artificial intelligence (AI) center of the hospital center of the University of Montreal (CHUM), is always ready to be enthusiastic when it comes to detailing the projects and achievements of his establishment in the use of artificial intelligence. But this speech therapist by training, who has plunged into the broth of new technologies, never departs from her mantra: all innovations must start from a need that must be assessed before considering making them permanent.
In the list established in 2021 by the American magazine Newsweek of the “the most innovative hospitals in the world”the CHUM was rewarded for its “innovative character” by the MTL Tech Awards as part of MTL Connecte, the festival organized by Printemps numérique. The CHUM prides itself on being the “the only healthcare establishment worldwide to have implemented, as of 2017, a globalized system where artificial intelligence has been introduced in all sectors of the hospital in order to accelerate innovation and transformation”explains its CEO, Fabrice Brunet.
This huge hospital, located in the heart of Montreal, with some 1,200 doctors, dentists and pharmacists, 4,000 nurses and 2,200 researchers, is also the first to have created, in 2018, a school of artificial intelligence in health which makes it possible to train young practitioners and spread knowledge.
The Covid-19 pandemic has given an undeniable boost if not to the projects – already abundant – of the innovation cluster, at least to their realization. In addition to a partnership with the Montreal start-up My Intelligent Machines, which made it possible to develop in record time a “monitoring” solution filtering the mass of scientific information linked to the new virus to help in the care of patients, the hospital wanted to very quickly improve the appointment booking system for other patients, especially those with cancer waiting for an X-ray or an operation.
“In September 2020, we started from a blank pagedetails Kathy Malas, but, as of April 2021, thanks to a partnership with two Quebec companies supported by our own teams of clinician researchers, we have managed to create, thanks to artificial intelligence, an appointment scheduling model that takes half the time to the agents in charge of this mission. »
The estimated financial gain of this optimization of the schedule of radiologists is 22,000 Canadian dollars per year (17,000 euros), but it is above all eleven additional hours of treatment per day which are thus freed up for patients, with the same human ressources. It is thanks to “data lake” clinical, administrative and research, supplied since 2017 at the CHUM and analyzed by data-scientistsnamely researchers who are also clinicians capable of judging the relevance of the information collected, that this new formula “outlook” diary could also see the light of day and be tested before being generalized.
“For direct patient care (…), AI never makes a decision in a closed loop, it is always a human being who decides in the end” – Kathy Malas, director of the innovation and AI cluster CHUM
AI has already made it possible to meet, in this hospital, a particularly sensitive challenge in oncology: being able to deliver the right treatment to a patient at the right time, each not reacting in the same way to the prescribed pharmacopoeia. Doctor Fred Saad, researcher at the CHUM and specialist in oncological urology, has notably developed an AI model to personalize the treatment of prostate cancer. Until then, the oncologist had to content himself with clinical data to establish the care protocol – the age of the patient and the misdiagnosis – genetic data and even imaging data related to his cancer were little used.
Personalized meal trays
AI has made it possible to combine all these elements and, by adding relevant socio-economic information, in particular the patient’s living environment, it now helps the practitioner to formulate an appropriate treatment for each patient. After several years of experimental phase, this predictive model is now considered as a “standard of practice”. But be careful, warns Kathy Malas, “For direct patient care, this model is all about helping the clinician. Artificial Intelligence never makes a decision in a closed loop, it is always a human being who decides in the end”.
If AI has made it possible to set up within the approximately 400,000 square meters that make up the CHUM, a system of self-guided vehicles that make 3,500 trips a day to transport medicines to the operating theaters and rooms or the necessary surgical instruments, it even interfered in the daily lives of patients. At each breakfast, lunch and dinner, the meal tray that is served to them has been concocted to respond to their pathology, the restrictions associated with it, but also to their preferences or lifestyles. Vegetarian, kosher, salt-free or protein, intended for a patient who has just had heart surgery or who is being treated in neurology, all menus are personalized.
Hundreds of projects
There again, for a year, it is the « machine », fed with thousands of data, which issues recommendations to the dietitian; he alone is responsible for signing the final menu. “It has been proven that the patient is happier to eat, so it contributes to his good recoverydetails Kathy Malas, and the nursing staff freed up time to take better care of the patient. Because this innovation makes it possible to meet the ambitions that we set for the use of Artificial Intelligence, it should quickly be generalized. »
Because, without positive feedback from patients and benefit attested and evaluated by hospital staff, certain experiments related to AI are purely and simply abandoned. But the CHUM continues to be full of projects; around a hundred today, at different stages, from a simple idea to the generalization of new practices. “How to make the supply chain, in equipment or drugs more efficient” or, “how to predict the time spent in the emergency room in order to set up an efficient prioritization system” are among the avenues explored.
On this last point, the challenge is considerable. Even at this cutting-edge Montreal hospital, ER occupancy rates regularly exceed 120%, with patients waiting for more than twenty-four hours on stretchers, who continue to count each day, by tens.
Devoted this year to the theme “Eco (systems) of hope”, MTL Connecte, whose The world is a partner, offers dozens of debates from October 15 to 23, 2022, in Montreal and in streaming. The event, organized by the Digital Spring, aims to approach the digital field in a transversal way, through its economic, social, cultural and environmental impacts in various sectors of activity. Among the sub-themes, artificial intelligence and this question: “How to guide this technology so that it becomes truly intelligent and improves human capacities?” “. Program and registration by following this link.