In a new study, researchers have developed a new computer program to help diagnose post-traumatic stress disorder (PTSD) in veterans by analyzing their voices.
They found that the artificial intelligence (AI) tool could help distinguish between the voices of those with or without PTSD. The accuracy was 89%.
The research was conducted by a team from NYU School of Medicine.
Research has shown that more than 70% of adults experience trauma at some point in their lives, and about 12% of people suffer from PTSD.
The main symptoms include strong, persistent distress when reminded of a triggering event.
Currently, a PTSD diagnosis is determined by clinical interview or a self-report assessment, but both inherently prone to biases.
It is important to develop a more objective, measurable, physical markers of PTSD.
In the current study, the team used a machine learning technique called random forests to solve the problem.
AI has the ability to “learn” how to classify individuals based on examples.
It can build “decision” rules and mathematical models that enable decision-making with higher accuracy.
The team first recorded standard diagnostic interviews of 53 Iraq and Afghanistan veterans who had military-service-related PTSD and 78 veterans who did not have the disease.
The recordings were then fed into voice software to generate 40,526 speech-based features captured in short spurts of talk.
The random forest program linked patterns of specific voice features with PTSD, including less clear speech and a lifeless, metallic tone.
These features had long been reported as symptoms of PTSD because the traumatic events change brain circuits that control emotion and muscle tone, which can affect a person’s voice.
The software could analyze words, in combination with frequency, rhythm, tone, and articulatory characteristics of speech.
This can help infer the state of the speaker, such as emotion, sentiment, cognition, health, mental health, and communication quality.
The team found that the program could distinguish between the voices of those with or without PTSD with 89% accuracy.
The team plans to train the AI voice tool with more data and further validate it on an independent sample. They hope to apply for government approval to use the tool clinically for PTSD.
The lead author of the study is Adam Brown, Ph.D., adjunct assistant professor in the Department of Psychiatry at NYU School of Medicine.
The study is published in the journal Depression and Anxiety.
Copyright © 2019 Knowridge Science Report. All rights reserved.