Michigan State University professor working on mobile solution to identify pills

Posted at 8:32 AM, Sep 14, 2016
and last updated 2016-09-14 10:42:40-04

As people get older, it’s not uncommon for their medicine cabinet to get a little fuller.

From blood pressure medication to pills for heartburn, people might have to take a garden variety of prescription drugs daily. But, if those pills ever get misplaced or mixed together, it can pose a risk to patients.

That’s why an educator at Michigan State University set out to find a solution.

Mi Zhang, an assistant professor in MSU's College of Engineering, and his team developed a mobile system to help patients identify pills.

"Our solution is based on a deep learning algorithm," said Zhang. 

The algorithm analyzes a photo taken from a smartphone and compares it to the pills from a database of 5,000. Even if the photo is taken from a weird angle or with bad lighting, the system can get a reading, according to Zhang. 

The system Zhang's team developed churns out a top 10 list of pills that starts with the most likely match. Zhang compares it to a "Google search that gives you a list of answers." 

He says that right now, as his system stands, there’s a 90 percent chance that the right pill will be in that calculated list of 10.

Zhang and his team were recently awarded $25,000 for winning first place in a U.S. National Library of Medicine competition that called for a pill recognition tool.

The next step, Zhang says, is to continue to improve on this solution. He says he wants to make it so that with 90 percent accuracy, the right pill will be the first one listed. 

They're also investigating how the technology can be applied in the future. 

"We are working with medical faculty at MSU to test it for different health applications," said Zhang.