(Reuters Health) - Voice analysis software can help detect post-traumatic stress disorder (PTSD) in veterans based on their speech, a study suggests.
Doctors have long understood that people with psychiatric disorders may speak differently than individuals who do not have mental health problems, researchers note in Depression and Anxiety. While some previous research points to the potential for distinct speech patterns among people with PTSD, it’s been unclear whether depression that often accompanies PTSD might explain the unique voice characteristics.
In the current study, voice analysis software detected which veterans had PTSD and which ones did not with 89 percent accuracy.
“Those with the PTSD talked more slowly (slower tongue movement), were more monotonous with fewer bursts of vocalization, were less animated and energetic (lifeless) in their speech, and had longer hesitations and a flatter tone,” said lead study author Dr. Charles Marmar, chair of psychiatry at NYU School of Medicine in New York City.
“Our findings suggest that speech-based characteristics can be used to diagnose this disease, and with further refinement and validation, may be employed in the clinic in the near future,” Marmar said by email.
Marmar’s team used an artificial intelligence program that “learns” how to classify individuals based on examples of speech.
First, researchers recorded hours-long interviews based on questions often asked by clinicians to diagnose PTSD. Altogether, they interviewed 53 Iraq and Afghanistan veterans with PTSD related to their service as well as 78 veterans without the disease.
Then, they fed the recordings into voice analysis software developed by Stanford Research Institute (SRI) International, designers of the “Siri” App, to yield a total of 40,526 speech-based features captured in short spurts of talk.
The software linked patterns of specific voice features with PTSD, including less clear speech and a lifeless, metallic tone, both of which had long been reported anecdotally as helpful in diagnosis.
While the study did not explore the disease mechanisms behind PTSD, the theory is that traumatic events change brain circuits that process emotion and muscle tone that affect a person’s voice, the study team writes.
The study was small, and it wasn’t designed to prove whether or how PTSD might directly cause changes in vocal patterns. It’s also possible that results might be different for people who experienced trauma unrelated to military service such as sexual assault or a natural disaster.
Other warning signs of PTSD may also be easier for family members to spot, said Dr. Ronald Pies of Tufts University School of Medicine in Boston.
“I think more general, observable indicators of trauma are more relevant in such cases,” Pies, who wasn’t involved in the study, said by email. “Noticing that a family member exposed to a recent trauma appears to be unusually irritable, aggressive, hyper-vigilant, or reports nightmares, flashbacks of the trauma, or appears socially withdrawn or depressed ... would warrant a clinical assessment.”
But it may not be too far in the future that a tool like the one tested in the study could be one way to identify people who need to be evaluated for PTSD, said U.S. Army Capt. Jeffrey Osgood of the Center for Military Psychiatry and Neuroscience at the Walter Reed Army Institute of Research.
“In a perfect world, I see this technology used as an early warning tool for PTSD,” Osgood, who wasn’t involved in the study, said by email.
It’s possible a version of the software tested in the study could be readily available, perhaps as a smartphone app, to analyze a person’s speech during and after highly stressful or traumatic experiences and to flag potential problems to patients or clinicians, Osgood said.
“This could prompt a more thorough screening and early intervention,” Osgood said. “However, more studies are needed before clinicians can confidently use this tool to help make diagnoses.”
SOURCE: bit.ly/2vhqhA0 Depression and Anxiety, online April 22, 2019.
Our Standards: The Thomson Reuters Trust Principles.