Smartphone sensor data to help fight bipolar
Smartphones are packed with a lot of sensors that the machine uses to understand its orientation and place in the world. For example, it knows when it is being held sideways, and can locate itself with GPS. But can those sensors also tell a doctor when you’re suffering from a mental illness?
Italian researchers have found that smartphones can be used to diagnose and manage mental health illnesses such as bipolar disorder, just by using their inbuilt sensors as a proxy for sudden mood changes.
Bipolar disorder is a condition characterized by extreme mood swings that veer from hyperactivity and elation, to severe depression and lethargy.
A study conducted by Venet Osmani at the Center for Research and Telecommunication Experimentation for Networked Communities (CREATE-NET) in Trento, Italy, indicates that a combination of smartphone sensors can detect when a user is having a manic or depressive episode.
In the study, Osmani gave 12 bipolar patients smartphones for four months and checked in on them every three weeks to monitor their conditions. It turns out that activity and location data on that smartphone could accurately predict a change in mood at the rate of 94 per cent. Monitoring phone calls for frequency and speed of calls bumped that accuracy up to 97 per cent.
Essentially, when users showed an increase in average activity (measured by GPS location and accelerometer speed), faster calls and more calls total, it indicated a shift into a manic episode. When activity, call length and call volume regressed, it signalled a shift into depression.
During this time, patients’ mental state was also determined by conventional methods (like counselling and evaluations) so the outcomes could be compared to the smartphone’s diagnosis.
Since these were detected before an actual bipolar episode, this type of measure could lead to early detection of an oncoming mood change, and timely treatment for the patient.
Dr Osmani told the Telegraph that his lab was also looking at measuring other mental disorders via smartphones. “We have carried out another study, investigating whether physical activity data can be used to detect chronic stress and burnout,” he said.
“On this basis, we have found that we can detect various stress levels with 71 per cent accuracy, solely based on the data provided by the accelerometer, while the subjects were holding the phone.”