Sunday February 05, 2017
The voice of PTSD diagnosis
Psychiatric disorders like post-traumatic stress disorder (PTSD) can be difficult to diagnose. Some patients are reluctant to disclose the full extent of their illness, and psychiatry can't rely on objective laboratory medicine to confirm a diagnosis.
Dr. Charles Marmar hopes to change that.
He's the chair of the Department of Psychiatry at New York University's Langone Medical Center and has devoted much of his career to the diagnosis and treatment of PTSD.
Charles believes that some diagnostic answers may be found in the human voice and is part of a research group that's using advanced machine learning to conduct vocal analysis to determine speech patterns that could indicate PTSD.
He defines PTSD as "a mental disorder that occurs in a minority of people who've been exposed to a highly stressful traumatic life experience, usually involving direct threat to a person's life, threat of severe injury, or witnessing very dangerous or gruesome events affecting other people we're closely connected to."
The symptoms of PTSD can be so severe they can interfere with a person's ability to work and carry on relationships.
The project is a collaboration with researchers at SRI International, a team that Marmar says, "has contributed to some of the most important technical advances in voice recognition."
He says the advantage of using advanced machine learning is that it allows researchers "to simultaneously investigate a very large number of predictors in the same individuals at the same time, and using advanced mathematical modeling, arrive at valid conclusions."
So far they have identified 40,000 uniquely different features of human speech among which, Charles says, are "about 200 which look promising for differentiating people currently suffering from PTSD." He hopes this will provide "a voiceprint analogous to a fingerprint."
Looking ahead, Charles hopes this work will lead to more objective tests to aid in the diagnosis of PTSD and other psychiatric disorders:
"We think in the future, a highly trained mental health clinician will utilize voice quality, blood tests, brain imaging tests, brain wave tests, and other features as diagnostic tests to refine and clarify diagnosis or select treatments."