by Jolynn Tumolo

By analyzing a patient’s spoken and written words, computer tools classified with up to 93% accuracy whether the person was suicidal, in a study published online in Suicide and Life-Threatening Behavior.

“While basic sciences provide the opportunity to understand biological markers related to suicide,” researchers wrote, “computer science provides opportunities to understand suicide thought markers.”

The study included 379 patients from emergency departments, inpatient centers, and outpatient centers at 3 sites. Researchers classified 130 of the patients as suicidal, 126 as mentally ill but not suicidal, and 123 as controls with neither mental illness nor suicidality.

Patients completed standardized behavioral rating scales and participated in semi-structured interviews. Five open-ended questions were used to stimulate conversation, including “Do you have hope?” “Are you angry?” and “Does it hurt emotionally?”

Using machine learning algorithms to analyze linguistic and acoustic characteristics in patients’ responses, computers were 93% accurate in classifying a person who was suicidal and 85% accurate in identifying whether a person was suicidal, had a mental illness but was not suicidal, or was neither.

“These computational approaches provide novel opportunities to apply technological innovations in suicide care and prevention, and it surely is needed,” said study lead author John Pestian, PhD, a professor in the divisions of biomedical informatics and psychiatry at Cincinnati Children’s Hospital Medical Center in Ohio.

“When you look around health care facilities, you see tremendous support from technology, but not so much for those who care for mental illness. Only now are our algorithms capable of supporting those caregivers. This methodology easily can be extended to schools, shelters, youth clubs, juvenile justice centers, and community centers, where earlier identification may help to reduce suicide attempts and deaths.”

References

Pestian JP, Sorter M, Connolly B, et al. A machine learning approach to identifying the thought markers of suicidal subjects: a prospective multicenter trial. Suicide and Life-Threatening Behavior. 2016 November 3;[Epub ahead of print].

Using a patient’s own words machine learning automatically identifies suicidal behavior [press release]. Cincinnati, OH: Cincinnati Children’s Hospital Medical Center; November 7, 2016.