New AI tool may help doctors pick the right weight loss, diabetes drug for patients

ATLANTA - Every day across north Georgia, tens of thousands of prescriptions are written and filled, as medical providers try to predict which medication is the best fit for their patients. 

One prescription many patients are asking for is for a weight-loss and diabetes injectable drug like Wegovy and Ozempic. 

New AI tool for weight-loss drugs 

Emory physician Dr. Caroline Collins teamed up with Emory Goizueta Business School professor Kaantirl Kuhnert to talk about how she decides whether these drugs, known as GLP-1 agonists, are a good fit for her patients. 

"I make this decision several times a day," Collins says. "So, it didn't take too long to really sit down, think of the variables, and then, use the AI software to help duplicate my decision." 

As part of a study, Collins, who is a board-certified internal and lifestyle medicine physician, talked Kuhnert through the variables that help her predict whether a GLP-1 agonist will be a successful fit for her patients. 

"Probably between 20 and 25 variables go into making this decision," Collins says.  "Some are more important than others, but all of them are important in some cases." 

"What we do is we come up with a set of variables that she really feels comfortable with," Kuhnert. "And then, what we do is we enter those, by the way, into the software." 

Wegovy is a semaglutide which aims to help slow down "gastric emptying," so that users feel full more quickly.

Drug-dispensing AI study 

As part of a study, Collins and Kuhnert used the AI software program the Tacit Object Modeler™ (or TOM) to create what they describe as a "decision-making digital twin," designed to help doctors predict whether this class of drugs will be successful for their patients. 

"It actually provides you with scenarios," Kuhnert says. "And it's learning from Dr. Collins what variables are most important to her." 

Collins says primary care or family practice physicians could use the software to help them predict whether their patient will benefit from the medication. 

"It's not really the computer making the decision," she says.  "It's the computer mimicking the decision I would make given these same variables." 

Kuhnert says AI could offer doctors a second opinion they can access in seconds on their phone. 

"It is a second opinion," he says. "We're not eliminating physicians or making it a physician's automatic decision but is a second opinion from an expert." 

Dr. Collins says she is happy to share her expertise in helping doctors make what can sometimes be a tough call. 

"These are really wonderful medications when used appropriately," Collins says. "But, when not used appropriately, they can hurt people."