Computer models predict suicide risk through health records. Suicide risk may soon be assessed by simply using medical records. A research team led by Boston Children’s Hospital has developed computer models that can screen electronic health records to identify patterns associated with suicide.
From 2000 to 2016, the rate of fatal suicides rose by 30 percent, and there were more than a million nonfatal suicide attempts in the year 2016 alone.
Suicide has now become the second most common cause of death among young people in the United States.
The computer models, which were developed using machine learning, can potentially alert health care professionals when a patient has been flagged.
One of the most significant findings is that the models are capable of recognizing suicide warning signs up to two years in advance of any suicide attempt.
“Computers cannot replace care teams in identifying mental health issues. But we feel that computers, if well designed, could identify high-risk patients who may currently be falling through the cracks, unnoticed by the health system,” explained said study co-senior author Dr. Ben Reis.
“We envision a system that could tell the doctor, ‘of all your patients, these three fall into a high-risk category. Take a few extra minutes to speak with them.'”
The investigation was focused on electronic health record data from more than 3.7 million patients ages 10 to 90 across five U.S. health care systems. Overall, the records contained a total of 39,162 suicide attempts.
The computer models were able to detect 38 percent of the at-risk patients with 90-percent accuracy. Incredibly, the cases were flagged in the health care records an average of 2.1 years before an actual suicide attempt.
The strongest predictors included drug poisoning, drug dependence, acute alcohol intoxication, and several mental health conditions. The researchers noted that other predictors were unexpected, such as rhabdomyolysis and HIV medications.
“There wasn’t one single predictor,” said Dr. Reis. “It is more of a gestalt or balance of evidence, a general signal that builds up over time.”
With funding from the National Institute of Mental Health, the team will now refine their computer models by incorporating more data, such as doctor’s clinical notes.
The study is published in the journal JAMA Network Open.
By Chrissy Sexton, Earth.com Staff Writer