A new artificial intelligence (AI) system created by a group of US researchers can assist medical professionals in identifying patients who are at risk of suicide.
AI-driven clinical alerts have the potential to enhance preventative initiatives in standard medical settings, according to researchers from the Vanderbilt University Medical Center.
The Vanderbilt Suicide Attempt and Ideation Likelihood model (VSAIL), an AI system, was tested by the team in three VUMC neurology clinics. Over a six-month period, they examined 7,732 patient visits for suicide risk during routine clinic visits.
The results, which were published in JAMA Network Open, demonstrated that the AI model could efficiently notify physicians of 596 screening alerts by evaluating routine data from electronic health records and estimating a patient's 30-day risk of attempting suicide.
The team also contrasted two methods: a more passive system that merely showed risk information in the patient's electronic chart, and automatic pop-up alerts that disrupted the doctor's workflow.
Doctors conducted suicide risk assessments in response to 42% of screening alerts, compared to 4% with the passive system, indicating that the interruptive alerts were significantly more effective.
According to the researchers, the new VSAIL can "help identify high-risk patients and prompt focused screening conversations," even though "universal screening isn't practical in every setting." They also suggested that similar systems could be tested in other medical settings.
The team discovered that interruptive alerts might lead to "alert fatigue" in physicians, even though they were more successful at triggering screenings.
