Bipolar disorder app predicts mood swings by eavesdropping on phone conversations
By Ben Coxworth
May 12, 2014
People afflicted with bipolar disorder must live with the fact that at any moment, they could launch into a major depressive or manic mental state. These mood swings can be so severe that dangerous, erratic behavior including suicide attempts can result. Researchers at the University of Michigan, however, are developing something that could prove to be very helpful. It's an Android app that listens to a patient's phone conversations, and detects the signs of oncoming mood swings in their voice.
Known as the PRIORI app, it constantly runs in the background on the patient's smartphone, recording all calls placed or received on the device, including weekly check-in calls with the patient's care team. By analyzing the patient's voice, it's able to identify subtle changes in their speech patterns, that are associated with the imminent onset of manic or depressive episodes.
Although the app is still in the developmental stage, the idea is that when such patterns are detected, both the patient and their caregivers will be alerted so that intervening and precautionary measures can be taken.
Due to privacy concerns, only the patient's side of conversations is recorded, and even that information is encrypted. Care teams only receive a report based on the processed data, and are not able to listen to the actual recorded audio itself.
In order for the app to work effectively, a baseline must first be set for each patient. This is part of the purpose of the check-in calls, at which times the patient's current speech patterns get matched to their reported mood. This means that the longer the app is used and the more familiar it becomes with the patient, the better it is able to predict their mood swings.
The researchers state that because other disorders such as schizophrenia and Parkinson’s disease likewise show up in the form of changes in speech patterns, PRIORI could conceivably also be used to predict the onset of those conditions.
Source: University of Michigan