Someone to Watch Over You
(and tell you to get some exercise)
In nearly every country in the world, life expectancy is rising. Embedded within this good news is the distressing increase of chronic diseases that occur later in life, perhaps most notably diabetes. While many researchers seek medical solutions to this problem, Center of Information Systems and Technology Professor Samir Chatterjee has been developing a technological one..
According to a report on aging by the United Nations, 10 percent of the world’s population was 60 years or older in 2000. By 2050, that number will rise to 21 percent. From 1985–2000 the number of people with type 2 diabetes—far more common than type 1 and typically diagnosed in adults 35 and older—had increased from 30 million to 285 million. Those two trends are dovetailing into an epidemic of elderly people who are managing a disease that requires constant diligence.
“This is a huge problem,” said Chatterjee. “In my research, I am targeting the older population, 65 and above. Everyone in that age group is at an advanced risk for dementia, which can obviously hinder proper management of type 2 diabetes.”
The American Diabetes Association has five cornerstones to self-management regimens. Some recommendations are relatively easy, such as keeping doctors appointments and measuring glucose levels daily. Others can get complicated or burdensome, such as taking medications at proper times, keeping a proper diet, and exercising regularly. Neglecting any of those guidelines, much less several, can have serious consequences for a diabetes patient: amputation, blindness, or fatal kidney disease.
To compel elderly people with diabetes to better manage their condition, Chatterjee drew on his background in persuasive technology, that is, technology that changes behavior or attitudes through persuasion. What he created is a subset of persuasive technology he calls persuasive sensing. With exploratory funding from the National Science Foundation, he installed a system of tiny sensors into the homes of two elderly individuals with diabetes that would measure physiological and biological parameters relevant to managing their condition.
One subject was a 60-year-old female, the other an 82-year-old male, who was also suffering from skin cancer. “Despite his ailments, he was amazingly active and energetic for someone his age,” Chatterjee said. “He was thrilled to be a participant, because he wanted to stay in his own home, and this monitoring is something that can help achieve that.”
In-home monitoring entailed that Chatterjee and his team, which includes CGU doctoral students Akshay Pottathil and Harry Xie, and postdoc fellow Jake Byun, to set up sensors in doors and medicine cabinets that could detect when they open and close; pressure pads in chairs, couches, and beds; and a monitor to record television use. The participants also wore BodyMedia armbands that measure steps taken as well as quality and quantity of sleep. Subjects were also instructed to report daily on what they had eaten and how much water they drank. The information from each device was transmitted back to Chatterjee’s lab.
“This was a typical before-and-after study,” he said. “After we wired up the homes, we collected data for a week, but had no communication with the subject. After that, we spent four weeks interacting with our subject, giving them feedback on their maintenance habits.”
Cell phone text messages provided specifically timed feedback to subjects three or four times a day. These updates were based on the subject’s behavior the previous 24 hours, and the information was presented with a combination of praise, encouragement, and concern.
“A message might say, ‘Yesterday you took 2,500 steps, but your goal is 4,000.’ So we give them a nudge,” said Chaterjee. “Or, ‘Yesterday you consumed 1,600 calories. Well done. Today you can have a cake.’ There are also messages that might alert the subject that they forgot to check their blood-glucose medication the day before. How would we know that? Because the medicine is in a cabinet, and we can detect that it wasn’t opened.”
At the end of every week a personalized health newsletter was created. A certified diabetes educator would visit the subjects and help explain the summaries, charts, and graphs. An additional benefit of these newsletters is that they can also be sent to a physician or family members.
When the five-week study concluded, results were encouraging enough to demonstrate great potential for persuasive sensing. The 82-year-old male’s behavior changed radically enough during the four-week period that it created striking health improvements. Previous to the study, his hemoglobin A1c levels—the average plasma glucose concentration over a prolonged period of time—fell from 12.9 to 6.6 percent. Generally, the goal for those with diabetes is to get that number down to 7 percent.
The second subject’s results were not quite as striking, though her hemoglobin A1c levels decreased from 8.9 to 8.5 percent. Still, parsing through the voluminous data recorded from this study, Chatterjee has discovered other benefits of persuasive sensing that go beyond behavior change.
One of the most exciting is the predictive-modeling. With only five weeks of data, Chatterjee and a subgroup of researchers were able to predict with 94 and 97 percent accuracy (for the male and female, respectively) the following day’s blood glucose value.
“I don’t know how doctors in clinics are currently monitoring this, but if we can expand the research—maybe take this to 50–100 homes—we might be able to really learn the value of our predictive model. And from there it would be relatively easy to send someone an alert that would pop up on their smartphone. Something like, ‘Your blood glucose values are predicted to be at this rate tomorrow, which is high. Consult your doctor,’” Chatterjee said.
He has already published some papers detailing the early results and potential of persuasive sensing. Judging from the positive feedback, Chatterjee sees many potential partners for follow-up research, including his own students at CGU. He was recently promoted to the Fletcher Jones Chair in Technology and Management, which will give him the ability to provide scholarships to two doctoral students who can assist him in his research. He has also received correspondence from colleagues in America and Australia about collaborating on future persuasive-sensing research.
“We are going to have to utilize technology, because when you look at the current demographics of diabetes, the current path we are on might be unsustainable without it,” said Chatterjee. “Thankfully, I really think we are on the cusp of a major breakthrough.”