Study: Speech recognition technology combined with AI could help detect PTSD

In a recent study published in the journal of Depression and Anxiety, a voice-analyzing AI was able to detect, with 89% accuracy, if a subject presented with PTSD.
By Laura Lovett
03:30 pm
Share

Voice-analyzing technology combined with artificial intelligence could offer a new way to detect post-traumatic stress disorder, according to a study published recently in the journal of Depression and Anxiety.

The research team found that the technology was able to use 18 speech features to identify subjects with the condition. 

Researchers said that this type of technology has the potential to make PTSD screenings easier and cheaper in the future. 

"Speech is an attractive candidate for use in an automated diagnostic system, perhaps as part of a future PTSD smartphone app, because it can be measured cheaply, remotely, and non-intrusively," lead author Adam Brown, PhD, adjunct assistant professor in the Department of Psychiatry at NYU School of Medicine, said in a statement.

TOP LINE DATA

The study, which included veterans of the wars in Iraq and Afghanistan, found that the AI technology was able detect if a subject had PTSD from vocal cues with 89% accuracy. 

The system also found that “monotonous speech, less change in tonality, and less activation” were indicators of the condition. 

METHODS 

The study included a group of 129 male veterans. All of the participants had been in Iraq or Afghanistan and were exposed to a war zone. Of those participants 53 were diagnosed with PTSD and 78 did not have the condition. 

Individuals with substance abuse disorder or a lifetime history of psychotic features, bipolar disorder, major depressive disorder and several other mental health conditions were excluded from the study. 

Subjects were interviewed by researchers using the standard Clinician-Administered PTSD Scale. Audio recordings of those interviews were fed into the AI software, yielding a total of 40,526 speech-based features captured in the recordings, which the AI program was able to then process. The AI algorithm eventually used 18 speech features to determine if a subject had the condition. 

It is worth noting the study was conducted using "extensive internal cross validation" — that is, the same sample was used for training and testing the algorithm. However, researchers noted that in the future "classifier endorsement requires a newly recruited external validation sample." 

THE BACKGROUND

There are a number of companies using speech patterns to detect for vocal biomarkers. Some of the players in the field include Sonde, which uses voice technology to monitor mental and physical health, WinterLight Labs, a system that can detect cognitive and mental disease from voice technologies, and BeyondVerbal, a startup that uses a speaker’s voice to give insights on health and wellbeing. 

Researchers noted that the technology is increasingly being looked at as a new avenue to detect mental health conditions. 

“There has been growing interest in speech-based techniques to screen for psychiatric disorders. Speech is an attractive candidate, as it can be measured at low-cost, remotely, non-invasively, and naturalistically,” authors of the study wrote. “Clinicians have long observed that individuals suffering from psychiatric disorders display changes in speech and routinely use impressions of voice quality as an element of mental status examination, including 'pressured' speech in bipolar disorder or 'monotone', 'lifeless', and 'metallic' speech in depression. More recently, automated techniques to analyze speech have been able to classify mood disorders on a number of speech features.”

IN CONCLUSION 

 “This study demonstrated that by using speech-based techniques, male Iraq and Afghanistan veterans with PTSD could be distinguished from warzone-exposed veterans without PTSD, neither of whom had [major depressive disorder] MDD,” authors of the study wrote. “The findings suggest that by combining frame-level short- duration and longer-duration prosodic features, high accuracy, sensitivity, and specificity for classifying PTSD can be achieved. The classifier assigns higher probabilities of PTSD to those with features indicating speech that is slower, more monotonous, and less change in tonality and activation.”

air max 90 essential amazon