New insulin pump

Intelligent Therapy
New software could provide guidance to Type 1 diabetics

by DWAYNE STEWARD
Features Writer
Perspectives
Monday Oct 08, 2007

New efforts by a computer scientist and an endocrinologist to design artificial intelligence software to manage insulin pump therapy could make life easier for patients with type 1 diabetes — and their doctors.

Dr. Frank Schwartz, an associate professor of endocrinology and director of the Appalachian Rural Health Institute’s Diabetes/Endocrine Center, is working with Cynthia Marling, an associate professor of electrical engineering and computer science, to create the smarter pump therapy.

The technology (U.S. patent pending), used in conjunction with newer glucose monitoring systems, will help patients manage their glucose, or blood sugar, levels. It will monitor daily glucose trends, remember previous patterns, and examine daily lifestyle information to make suggestions to physicians about what patients can do to maintain good glucose control.

The most advanced insulin pumps currently available have glucose sensors that measure the glucose level continuously, which is a tremendous advance over previous technology. However, they do not adjust insulin levels automatically or make suggestions to help people with diabetes. The patient or physician must determine what adjustments in insulin pump doses should be made to correct any problems.

In the Ohio University study, five graduate research assistants shadowed Dr. Schwartz and Dr. Jay Shubrook, an Ohio University assistant professor in family medicine and trained diabetologist, as they identified and diagnosed problems with their patients.

Patients documented every detail of their treatment in an online database for six weeks. The physicians reviewed the data for each patient and then suggested an individualized course of action. The program is being taught to recognize common patterns of glucose problems (either too high or too low) as well as possible solutions for these problems.

“For example, if someone’s glucose levels are high at night, Dr. Schwartz may tell them they need to increase their pump basal rates of insulin infusion,” Marling says.

If the recommendation works, it is plugged into the computer program as a solution to a specific problem.

“We’ve identified 50 problems and solutions by working with 20 patients so far,” she says. “If we study 50 patients, we’d have a practical tool we could let patients use.”

In the future, this type of artificial intelligence will help patients identify problems and offer treatment suggestions without physician intervention, says Marling, who previously developed artificial intelligence software for physicians who treat patients with Alzheimer’s disease.

The project receives funding from the Ohio University College of Osteopathic Medicine, the Russ College of Engineering and Technology, and Medtronic MiniMed, an insulin pump manufacturer, which provided $5,000 in equipment.

“We’re also in the process of registering the patent for this software and meeting with the insulin pump companies in hopes of forming a collaborative partnership with one of them,” Schwartz says.

This would not only bring in the private funding needed to continue the research, he says, but would put the technology in the hands of a national manufacturer that would help it reach more patients.