Tech & Health

February 11, 2021

Study will delve into EHR for signs of suicidality

Retrieval of clearer, more complete information from the EHR could go a long way toward improving predictive models of who will next be at risk of suicide, thereby improving care for patients with suicidal thoughts and behaviors.

 

by Paul Govern

When a patient attempts suicide or reports suicidal thoughts, the trail this leaves in the electronic health record (EHR) is often fuzzy.

For one thing, diagnostic codes alone won’t establish whether someone had a suicide attempt last week or 10 years ago. Retrieval of clearer, more complete information from the EHR could go a long way toward improving predictive models of who will next be at risk of suicide, thereby improving care for patients with suicidal thoughts and behaviors.

Colin Walsh, MD, MA

With the aid of a $1 million, two-year contract from the U.S. Food and Drug Administration’s Sentinel Innovation Center, Colin Walsh, MD, MA, and colleagues at Vanderbilt University Medical Center will develop natural language processing (NLP) to hone in on any text entered in the EHR that touches on patients’ suicidal thinking and behavior.

“Diagnostic coding by the health care team heavily under-captures suicide attempts and identifying the timing and circumstances of these events from patient charts is especially challenging,” said Walsh, assistant professor of Biomedical Informatics. “With improved NLP of care team notes and messages in the EHR, we hope to better identify suicidal ideation and suicide attempts when they actually occurred. The result will help us better understand the scope of these problems while designing better care programs to prevent them, enabled by the machine learning algorithms under development by our team.”

Walsh and colleagues will also test the ability of NLP to capture medication-related neuropsychiatric events and enable better medication safety monitoring for the FDA’s Sentinel Initiative.

The FDA contract complements an ongoing National Institutes of Health grant to Walsh and colleagues for development of artificial intelligence to explore suicide risk.

From 1999 to 2017, the all-ages suicide rate in the United States increased 33%, from 10.5 to 14.0 per 100,000 population. In 2017 there were 47,173 recorded suicides, making it the nation’s 10th leading cause of death.

If you’re having thoughts of suicide, call the National Suicide Prevention Hotline, 800-273-8255, or text HOME to 741741, the Crisis Text Line.