Artificial intelligence tools speed up the process of identifying people who inject drugs

Artificial intelligence tools speed up the process of identifying people who inject drugs

RESULTS

An automated process combining natural language processing and machine learning identifies people who inject drugs (PWID) in electronic health records faster and more accurately than current methods that rely on manual record reviews.

BACKGROUND

Currently, people who inject drugs are identified using International Classification of Diseases (ICD) codes provided in patients' electronic health records by healthcare providers or extracted from those notes by trained human coders who review them for billing purposes. However, there is no specific ICD code for intravenous drug use, so providers and coders must rely on a combination of nonspecific codes as proxies to identify PWIDs—a slow approach that can lead to inaccuracies.

METHOD

Researchers manually reviewed 1,000 records from 2003 to 2014 of people admitted to Veterans Administration hospitals with Staphylococcus aureus bacteremia, a common infection that occurs when the bacteria enter openings in the skin, such as injection sites. They then developed and trained algorithms using natural language processing and machine learning and compared them to 11 proxy combinations of ICD codes to identify PWIDs.

Limitations of the study include potentially poor documentation by providers. Additionally, the dataset used is from 2003 to 2014, but the injection drug use epidemic has since shifted from prescription opioids and heroin to synthetic opioids such as fentanyl, which the algorithm may miss because the dataset it learned the classification on does not contain many examples of this drug. Finally, the results may not be generalizable to other circumstances because they are based entirely on Veterans Administration data.

IMPACT

The use of this artificial intelligence model significantly accelerates the process of identifying PWIDs, which could improve clinical decision-making, healthcare research, and administrative monitoring.

COMMENT

"Using natural language processing and machine learning, we were able to identify people who inject drugs in thousands of notes within minutes, compared to the several weeks it would take a manual reviewer," said lead author Dr. David Goodman-Meza, assistant professor of medicine in the Division of Infectious Diseases at the David Geffen School of Medicine at UCLA. “This would allow health systems to identify PWIDs to better allocate resources such as syringe service programs and substance use and mental health treatment for people who use drugs.”

AUTHORS

The other researchers on the study are Dr. Amber Tang, Dr. Matthew Bidwell Goetz, Steven Shoptaw and Alex Bui of UCLA; Dr. Michihiko Goto of the University of Iowa and Iowa City Medical Center, VA; Dr. Babak Aryanfar of the VA Greater Los Angeles Healthcare System; Sergio Vazquez of Dartmouth College; and Dr. Adam Gordon of the University of Utah and the VA Salt Lake City Health Care System. Goodman-Meza and Goetz also have appointments at the VA Greater Los Angeles Healthcare System.

DIARY

The study was published in the journal Open Forum Infectious Diseases.

FINANCING

The US National Institute on Drug Abuse funded this study.

Source:

University of California, Los Angeles (UCLA), Health Sciences

.