An application that uses artificial intelligence to encourage users to maintain healthy lifestyles: this is what a team of researchers from Concordia University and the École de technologie supérieure is developing.
“Imagine someone who follows a program of fitness through physical activity, illustrates Simon Bacon, professor at Concordia University and co-holder of the Chair in artificial intelligence and digital health for behavior change in health. Each week this person needs to make a few more changes, achieving goals that they have set for themselves. By week eight, she had a very difficult week at work, which disrupted her schedule and plans. “
When she gets home, the app she uses to track her progress asks her for updates. She then responds, by video, “it’s okay, it’s okay”. By analyzing her words, her defeatist tone and even her shrug, the app understands that the user has a decrease in motivation. She then replies: “Sometimes changes in habits can be difficult, and obstacles can stand in your way. “
“A simple manifestation of understanding can have a very positive effect,” says Bacon.
According to him, the majority of people want to move more and eat healthier, but are likely to shrink from problems such as lack of time or the distance between their home and their favorite sports center.
Mr. Bacon defines this hesitation as ambivalence, a crucial state that can derail the process. “By combating people’s ambivalence, we can really help them stay committed to good lifestyles,” he adds.
Address the gaps in the system
The health system offers support to people who have developed significant chronic diseases related to their lifestyle. But for the majority of people, there are very few services available.
“In the health system, we can offer you solutions, but we do not support you in the process. However, many people would benefit from a few small changes to be in better health ”, estimates the researcher.
This is where the application of Mr. Bacon and his team, made up of a dozen researchers and students, comes into play. Its current version, in English and French, allows you to choose weekly goals and record your progress. There are also virtual meetings with real professionals along the way. They can detect participants’ hesitations and advise changes to the game plan accordingly. However, given the low availability of professional resources, this solution cannot be accessible on a large scale.
Mr. Bacon’s team is therefore working to replace these stakeholders with artificial intelligence. To do this, the first step is to detect ambivalence in the user.
Behavioral health researchers analyze hours of videos from dozens of volunteer users, annotating them when they notice clues of ambivalence, such as facial expressions, gestures and intonations that contrast with their words. Computer scientists and computer engineers then use this data to develop an algorithm that can detect the ambivalence of a multitude of individuals in a variety of situations. This will make it possible to generate an intervention adapted to their responses.
Over the next year, registrations of 200 to 300 users will be used to refine the new version of the application. The latter should be available in 2023, under a still unknown name. It could be used in health systems, but also individually by a large audience, says Bacon.
For anyone who hasn’t gotten to exercise for more than two weeks in a row after January, there is still hope. In the meantime, the expert recommends the strategy of small steps: to make modest and gradual changes, but realistic.
This content is produced in collaboration with Concordia University.