Berkeley Lab researchers use AI to determine veterans' suicide risk

Lawrence Berkeley (Calif.) National Laboratory researchers are applying artificial intelligence and analytics to EHR data to help the Department of Veterans Affairs identify veterans' risk of suicide.

The Berkeley Lab team aims to develop patient-specific algorithms that can provide tailored suicide risk scores, such as whether an individual who has previously been admitted to the hospital for a suicide attempt will make another attempt within 30 days. The information would then be made available to the VA, caregivers and patients.

The research stems from a previous research project in 2018, which Berkeley Lab students applied deep learning methods to a dataset from a Boston-based hospital intensive care unit. Using EHR information from an estimated 40,000 patients, the students created algorithms to perform statistical analysis and identify key factors related to suicide risks.

To determine key factors that lead to suicide, researchers had to separate structured and unstructured EHR data. To do this, one research student built a deep learning network that can use unstructured data such as discharge and physician's notes to distinguish and classify individuals at risk for suicide.

"We believe that, for suicide prevention, the unstructured data will give us another side of the story that is extremely important for predicting risk — things like what the person is feeling, social isolation, homelessness, lack of sleep, pain and incarceration," Xinlian Liu, PhD, associate professor of computer science at Frederick, Md.-based Hood College who participated in Berkeley Lab's 2018 program, said in a news release. "This kind of data is more complicated and heterogeneous."

The Berkeley Lab team will continue its suicide-risk research this summer. The research is part of the VA and Department of Energy's partnership that focuses on suicide prevention, prostate cancer and cardiovascular disease. Through the collaboration, the departments aim to work with national laboratories to apply high-performance computing, software development and networking resources to medical datasets collected by the VA. The datasets comprise information from an estimated 700,000 veterans and EHR data from 22 million veterans.

More articles on artificial intelligence:
Google pulls plug on AI, tech advisory board
Top AI Google employee jumps ship to Apple
AI may be able to determine success of IVF embryo implantations, study shows

© Copyright ASC COMMUNICATIONS 2020. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.


Featured Webinars

Featured Whitepapers