Watson-Powered Diabetes App Predicts Hypoglycemia

January 08, 2016

LAS VEGAS — Watson, the IBM supercomputer that won TV's Jeopardy, will soon be able to help patients with diabetes prevent hypoglycemia, medical device maker Medtronic announced here at the giant Consumer Technology Association Digital Health Summit.

The company expects to introduce a smartphone app this summer that will provide timely hypoglycemic warnings to patients using its insulin-management devices.

The idea for the app arose from an unpublished study that Medtronic conducted using Watson's "cognitive computing" — buzz words heard commonly at the show. The supercomputer analyzes high volumes of data, detects patterns, understands natural language, continuously learns, and simulates human reasoning. Watson is already providing oncologists with evidence-based treatment options, and IBM plans to put the computer's optical recognition capabilities to work in radiology.

Watson crunched data from Medtronic's MiniMed insulin pump and continuous glucose monitor, along with carbohydrate intake, and was able to retrospectively predict hypoglycemic events that the 600 study participants experienced 3 hours in advance with 80% accuracy.

A smartphone alert powered by Watson would give patients the chance to prevent a hypoglycemic episode by drinking orange juice or eating a banana, said Annette Brüls, vice president and general manager of Medtronic's global diabetes unit.

Hours Ahead

"That's the beauty of it," Brüls told Medscape Medical News. "With the alert 3 hours in advance, there's time to intervene. This is self-management."

The app isn't the company's first self-management tool for patients with diabetes. Last September, Medtronic introduced an app called MiniMed Connect, which lets patients view data from the company's MiniMed insulin pump and continuous glucose monitor on a smartphone.

Brüls said that MiniMed Connect warns patients 30 minutes before their blood glucose levels reach preset high or low limits, but by that time, they're already getting into some trouble. MiniMed is designed to suspend insulin delivery during a hypoglycemic run-up. There is no need for that with the new app's 3-hour alert, said Brüls. "Nothing has happened yet."

Brüls explained that Medtronic's collaboration with IBM and its supercomputer will make its diabetes-management apps even smarter in the future.

Personal Diabetes Coach on a Smartphone

With Watson making sense of a patient's glucose levels, insulin intake, diet, physical activity, and other data, the software can serve as a patient's personal diabetes coach, offering a wide range of lifestyle recommendations on how to stay on an even keel.

In the future, it might be able to explain why a patient's blood glucose level spiked or identify meals that are best for glycemic control.

The challenge for companies such as Medtronic is to feed Watson more and more pertinent data without requiring so much data entry that the patients fingers give up.

The 600 patients in the Medtronic study had to input their carbohydrate consumption daily. Brüls said that new information technology — like some of those exhibited here at the Consumer Technology Association Digital Health Summit — should make that job easier as time goes on. Restaurant menu items and their dietary data can be synched to a patient's diabetes app, for example.

In a panel discussion on cognitive computing, Brüls said that digital healthcare can never have too much information.

"As long as you can digest it and make sense of it," she said, "it's not a problem."

Consumer Technology Association (CES) 2016 Digital Health Summit. Presented January 8, 2016.

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as:

processing....