UT researchers create AI system to improve nutrition

Rylie Lillibridge, Senior News Reporter

UT researchers developed an artificial intelligence system called Nutri to help healthcare providers discuss nutritional goals with their patients. 

Marissa Burgermaster, who developed the idea for Nutri, said that though patients and providers recognize the importance of nutrition, the little nutritional training that doctors receive combined with short appointments makes it hard for effective conversations about nutrition to occur.

“Nutri takes some of the burden off of the patients and providers by doing some of the hard work, the stuff that humans aren’t necessarily really good at,” said Burgermaster, assistant professor of  nutritional sciences and population health. “It highlights the changes that are expected to have the biggest impact for a patient and then lets the doctor and the patient do what they’re best at – talk about how to make it work in their life.”

Patients input their diet information into Nutri by filling out the Automated Self-Administered 24-hour Dietary Assessment Tool, which is processed by Nutri and compared to the official Dietary Guidelines for Americans. The processed data is presented in an interface and then used during a patient’s appointment. 

“Goals are presented to patients and providers in a prioritized order, and it walks them through a workflow that facilitates a conversation about (their nutritional goals),” said Madalyn Rosenthal, research program coordinator for Dr. Burgermaster’s lab.

At the end of an appointment, the information gets summarized for providers in an automated clinical note and listed in a handout for the patient.

“One thing that’s really cool about Nutri is it’s designed for two different people: … patients and providers,” said Rosenthal, a nutritional sciences doctoral student. “It’s not only designed for those two people, but it’s designed to help them use it together.”

Burgermaster said collaborative diet goal-setting in Nutri can help patients achieve higher behavioral intention, which means a stronger likelihood of following a behavior as well as increased confidence that they can actually achieve the behavior.

“If we can use Nutri to improve behavioral intention and self-efficacy, it will then have an effect on diet change and ultimately on the clinical markers that are related to diet and chronic disease,” Burgermaster said.

The researchers are still analyzing data from a recent Nutri trial at Lone Star Circle of Care Austin, but Burgermaster said the primary care providers used Nutri in 100% of the appointments at which it was presented to the provider. 

“We’re just so proud that 100% of the time the providers had the opportunity to use Nutri, they did,” Burgermaster said. “They went through the three steps and set a goal with their patient.”

Burgermaster said the Nutri team is in the process of planning a larger-scale trial and working on features such as tailored food boxes that will reduce the barriers between patients and their goals.

“This is a really unique approach to addressing nutrition in an environment (where) people want to talk about it but maybe don’t have the resources to do so,” Rosenthal said. “I’m very proud of the user testing that we did to make sure that … providers and patients alike have something that’s going to work for them and that they want to use.”