Mobile phones are the modern American’s most faithful companion. They follow us from home to work, the gym and grocery store, and back again. They never forget a birthday, anniversary or soccer game. And they are always available to offer advice about our finances, spelling and love life — no matter what time of day or night.
It isn’t a theory but a fact to say that your smartphone knows more about you than you do.
A growing number of scientists are starting to mine this data in the hopes that it will help them understand what makes you happy or sad, and pinpoint signs of a disease long before it can be diagnosed by a blood sample or MRI, helping you live longer and better.
In one of the first of a number of studies in the works to be published, researchers at Northwestern University Feinberg School of Medicine believe they have found a way for your smartphone to determine if you’re depressed.
Through an ad on Craigslist, the research team recruited 28 volunteers — 20 females and eight males ranging in ages from 19 to 58 — and collected GPS and phone usage information on them for two weeks. The data was collected by an app called “Purple Robot” that was developed in-house. The researchers also asked the volunteers to complete a number of health questionnaires. It turned out that exactly half had some signs of depression.
In a study published Wednesday in the Journal of Medical Research, they report that the more time you spend using your phone the more likely it is that you’re depressed. That link didn’t hold true for 100 percent of people, however. A second analysis that looked at how people move through time and space showed stronger correlations.
By using this data the researchers were able to identify people with depressive symptoms with a startling 87 percent accuracy, although they noted that the results are based on a small sample size and are therefore preliminary.
“If these methods are successful in finding out if someone has depression, symptoms won’t require any input from the patient. We’ll be able to passively and objectively measure behavior without a patient having to report this every day,” said lead author Sohrob Saeb, a computer scientist.
Using some pretty complex algorithms and mapping, Saeb and his colleagues found that three ways of looking at how a person moves seem to impact the presence and severity of depressive symptoms:
• Circadian movement.
• Normalized entropy.
• Location variance.
Saeb defined circadian movement — a term his team made up based on the idea of the 24-hour circadian rhythm of some animals and plants — as how regularly people are moving between locations from day to day. “If they move from home to work at the same time across days or at different times,” he explained in an interview. They assigned scores based on how “regular” their movements were according to these measures. The highest possible value would go to someone who went to exactly the same place at exactly same time every day. “That kind of person didn’t exist in our study,” Saeb said. The higher the score, the less likely a person was to have depressive symptoms.
Normalized entropy is a measure of how uniformly you distribute your time across different locations. If a person’s entropy score is zero, “you are always staying in the same location at the same time,” Saeb explained. “At the other end you are spending time equally in different locations.” The higher the score, the less likely a person was to have depressive symptoms.
The last measure — location variance — Saeb defined as “mobility in space, how much you are moving.” So if a person moved a lot in terms of physical distance, they got a higher score. Again, the higher the number, the fewer the depressive symptoms. “That was not very surprising,” Saeb said.
The implications of this type of work are enormous, not only for the future of health care but for people’s privacy. It wasn’t so long ago that people worried about the data collected by marketers using information about your purchasing habits from your credit card, marketing surveys and ZIP code. That amount of information seems very limited now in comparison to the intimate moments and habits your cellphone is logging every second.
Saeb said the group’s next step is to try to duplicate the study in a larger population and to add more sensors so they can measure the types of physical activity, sleep, communications, and other aspects of a person’s life. One promising area, he believes, is in looking at speech patterns and what they can tell about your mental health.
