Bipolar Speech Patterns: A Different Kind of Subtext

Thursday, April 30, 2015
Category: 
Specialty: 

Researchers are developing smartphone applications that may predict manic or depressive episodes based on speech patterns.

Although pharmaco- and psychotherapeutic strategies can manage the emotional swings caused by bipolar disorder, the disorder’s unpredictability limits these approaches’ efficacy. In response, researchers are building upon studies that link acoustic changes to mood states. Their goal is to develop apps that identify precursors of manic and depressive episodes.

A clinical trial whose results were published in 2012 in Biological Psychiatry identified vocal acoustic patterns that significantly correlated with depression severity, including pause time, speech/pause ratio and speaking rate.

Vocal Cues

Researchers at the University of Michigan are testing an app that analyzes speech patterns to detect subtle variations indicating mood changes. Dubbed PRIORI, the app functions in the phone’s background and records the patient’s side of any call. Recordings are stored in compliance with patient privacy laws.

Weekly meetings with a care team establish a baseline for the patient’s mood. Because the sessions are recorded, the team can identify acoustic qualities in the patient’s speech and link those to the mood he or she is in.

Changes — such as long pauses — in vocal acoustics may indicate depressive states, and rapid topic changes and fast talking suggest manic swings. The researchers hope to discover precursors to these shifts so patients know when they need to get help.

“You can get a good idea about their state of health in just one to two sentences,” says Melvin McInnis, MD, Professor of Psychiatry at the University of Michigan, who is involved with the PRIORI research. “Likewise, a computer can analyze acoustic features of speech and identify a particular range of how a person is talking. We anticipate that we’ll be able to find and extract acoustic patterns that will be most predictive of mood state changes and [use these to monitor patients].”

Improved Data, Expanded Applications

PRIORI represents a potentially marked improvement over currently available smartphone applications. With existing apps, patients record data in a journal-like format, enabling them to identify events that might trigger mood swings, as well as remember medication instructions and keep track of questions they want to ask their physician or psychologist. Yet these apps depend on patient compliance, whereas data collection is automatic with PRIORI.

University of Michigan researchers hope to apply their findings to other mental health disorders, such as schizophrenia, as well as to general health conditions.

“The bulk of illnesses [that affect organ systems above the diaphragm] affect the way an individual speaks,” Dr. McInnis says. “We’re looking into … whether you can use the acoustics of speech to be able to predict when someone is on a trajectory for a health-state change, be it a mood change or any other state of health.”