An attempt to detect the timing of manic-depressive patients' emotional instability by "voice" is in progress

ByAirman Magazine

Bipolar disorderBipolar disorder) The patient became "depressed state" in which the mood became abnormally elevated, irritated and became angry, becoming "manic state", depressed mood, decline of motivation, interest, mental activity, insomnia, etc. became conspicuous As the daily mood changes with the corner, it seems that there are occasions when you get lost in work or collapsed human relations because of its influence. A project is underway to make it possible to confirm the "mood" cycle of such bipolar disorder patients from smartphones.

Mood Ring - Cell Phones Can Hear Depression in People 's Voices - Facts So Romantic - Nautilus

A team of psychologists led by Melvin McInnis, a researcher at the University of Michigan, is a group of psychologists who sense bipolar disorder patients in emotional instabilityPRIORI"We are promoting a project called.

In the project, we will develop applications to find faint signs when bipolar patients become manic or depressed by collecting data when users are talking using mobile phones going. Once this application is completed, it will be possible for the patient to be aware of the fact that the mood is changing instinctively in emotional instability, so recognize that medical assistance is needed before losing consciousness It also leads to a reduction in the risk that a fatal mistake will be committed in work and human relations due to emotional instability.

It is, of course, possible to read the mood of a person who does not suffer from a mood disorder such as bipolar disorder, and also how it felt. However, especially from the voice of people with mood disorder, changes in emotion are easy to read. For example, a person in a depressed state, at a relaxed pace, has a long breath between words and talks in a monotonous and hoarse voice, or in a manic state as if it had been overwhelmed with a loud and fast pace There are features such as speaking.

The research team is extracting some features from the bipolar disorder's voice to identify mood deterioration and so on, in order to ascertain whether antidepressants are working firmly on patients It will also be an indicator of.


Emily Mower Provost and Zahi Karam who are researching computer science at the University of Michigan are developing applications for mobile phones. By analyzing the size and tempo of the voice, we are working on the development of a system that can inform bipolar patient about it before it becomes emotional instability.

For example, "the size of voice" is an important indicator to confirm that it is a manic state or a depressed state, but see information on other voices and "When did you get emotional instability?" The accuracy of detecting timing when this emotional instability becomes a higher level becomes higher. However, it is not clear at the moment what kind of data is useful for detecting the most change in mood. However, it seems that there is already an algorithm for estimating the state of bipolar patient from voice.

The following graph is a graph of data outputted by an algorithm for estimating the state of a patient from the voice of a patient with bipolar disorder. The waveform on the left is the waveform of the voice of the bipolar disorder patient (1 second) when manic, normal (Euthymic), depressed (depressed), the right side shows the "mood level" that the application can detect . The colorful rectangle beside the wavy line shows various features extracted from the voice, for example, "1" and "9" at normal times, "5" and "10" at depression are larger than others You can see that it is changing.

Currently, it seems that the application is being verified for 10 bipolar patients. The number of subjects will be expanded to 60 people in the future, subjects will be monitored using mobile phones for several years.

in Science, Posted by logu_ii