Doctors diagnose mental illness very differently from the way they diagnose other types of disease. For a physical illness, doctors run a series of tests, examining the part of the body that is sick or damaged for indicators called biomarkers. But to tell if someone has a mental illness, clinicians must rely on interviewing their patients and interpreting the symptoms they describe.
“If you came into the emergency department with chest pain, you would not want to have cardiac surgery without biomarkers to confirm that you had heart disease,” said Dr. Charles Marmar, chairman of the department of psychiatry at New York University's Langone Medical Center, in an interview with Healthline.
But thanks in part to Marmar, the field of mental-illness diagnosis could soon change.
Marmar is leading a new study to find biomarkers for post-traumatic stress disorder (PTSD), which affects 7.7 million Americans. PTSD occurs after someone is exposed to a traumatic event, such as a sexual assault or the loss of a loved one, and doesn’t recover within a few months. Today, clinicians diagnose PTSD by asking whether a person has symptoms —which include flashbacks, nightmares, panic attacks, insomnia, and feelings of emotional numbness or detachment. Marmar’s team is instead seeking concrete medical ways to diagnose PTSD.
His study will examine the body on many different levels, looking for structural and functional differences in the brain, genes, hormones, and proteins of PTSD sufferers, as well as one other telltale factor: speech.
Tell Me What the Problem Is
Marmar is collaborating with scientists at SRI International, a nonprofit research organization, to analyze patients' speech patterns for signs of PTSD. “Both the content and the form of speech can be the source of biomarkers for a certain state,” said Dimitra Vergyri, assistant director at SRI’s Speech Technology and Research Laboratory, in a podcast. “The content refers to the actual spoken words. The form involves the acoustic characteristics that describe how you say things. Also the rhythm, how fast one speaks, the pauses between the words—all these indicate emotional state or even stress levels.”
Among the data that Marmar is collecting are patient interviews, using high-quality voice recordings with the subjects speaking. Currently, Vergyri is examining 20 people with PTSD and 20 people without PTSD to find the differences between the two groups.
“We’re currently getting promising results, where we can predict the subject’s condition significantly better than chance, but we still need a lot of work and need to analyze much more data before we can understand how well the system can do or why it is even working,” she said. Past work by SRI has involved analyzing speech for identification purposes and for biomarkers of depression.
Rachel Yehuda, a professor of psychiatry at the Icahn School of Medicine at Mount Sinai and director of mental health at the James J. Peters Veterans Affairs Medical Center in the Bronx, thinks that blood- and brain-imaging biomarkers will be the most useful for diagnosing PTSD. However, she thinks speech analysis could also be useful. “Any avenue should be explored,” she said in an interview with Healthline. “Speech has a unique signature that could also be captured with other markers, such as blood biomarkers, identifying a ‘unique’ fingerprint.”
Speech also has the advantage of being remotely accessible, Vergyri said. Voice data can be gathered from soldiers on the battlefield and easily transmitted back home for doctors to analyze.
The Trouble with Self-Reporting
For now, doctors have to make do with patients’ self-reports. And self-reporting is particularly difficult for PTSD. “The problem is, at least among war fighters—and to some extent police officers, firefighters, and civilians—there’s a great deal of understanding of subjective distress related to trauma,” explained Marmar. “War fighters are in a very macho culture, and it’s difficult for them to acknowledge to themselves and others that they’re suffering from stress-related problems.”
Patients might also keep their PTSD hidden to protect their careers or security clearance, which can be revoked upon diagnosis with a mental illness. Meanwhile, other patients might overstate their symptoms, trying to get compensation from the government. “The limitation is that if a patient wishes not to disclose information, the clinician might not pick it up,” said Yehuda. “If a patient wishes to give the impression of symptoms not present, the clinician may fail to recognize this as well.”
Yehuda recommends caution, however, in using biomarkers alone. “We want to make sure that we do not increase stigma for patients,” she said. “The absence of a marker may invalidate suffering, and that would not be a positive development. We want to make sure that biomarkers have positive impacts on treatment and diagnosis while avoiding interpreting biomarkers as signs of permanent disability or illness, particularly in the case where reversal of a biological situation is possible. In many cases, PTSD can go into remission with total symptom alleviation.”
Still, discovering biomarkers is an important first step toward treating PTSD as the medical disorder it is. “The big problem with post-traumatic stress is that we have no objective way to know for certain if someone really does or does not have the diagnosis, and until we have that we won’t know for sure who should be given treatment,” said Marmar.
“We don’t have biomarkers which tell us which kind of treatment to give somebody: psychotherapy, medication therapy, brain-stimulation therapy, and others which are available. We don’t have biomarkers to know when someone’s definitely recovered from post-traumatic stress disorder or not. We don’t have biomarkers to determine who’s going to have a good course and recover with time, and who will have a difficult course and need intensive treatment," Marmar added. "So there is an urgent need to develop objective markers.”