What's Happening?
Researchers at the University of New Mexico School of Medicine have conducted a study analyzing electronic health records of over 1.3 million patients from the Veterans Health Administration (VHA). The study, published in the Journal of Medical Internet
Research, highlights a significant gap in how health systems track self-harm. It found that diagnosis codes captured only about one-fourth of clinically documented self-harm history. The research utilized a machine learning method called PULSNAR to estimate the presence of self-harm history that was not visible through diagnosis codes. The study revealed that documented self-harm was present in about 7.9% of patients, compared to the 1.85% visible through diagnosis codes alone. This discrepancy underscores the challenges in accurately tracking mental health conditions, which can affect clinical awareness and planning for mental health services.
Why It's Important?
The findings of this study are crucial for improving mental health services for veterans. By identifying gaps in the documentation of self-harm, the research highlights the potential for underestimating the need for mental health services. This can have significant implications for public health policy and resource allocation within the VHA. Accurate tracking of self-harm is essential for effective mental health care, as past self-harm is a strong predictor of future self-harm and suicide risk. The study's use of AI to uncover these gaps could lead to better planning and delivery of mental health services, ultimately benefiting veterans who rely on the VHA for care.
What's Next?
The research team suggests that the PULSNAR method could complement existing VHA mental health and suicide-prevention efforts by providing a scalable way to measure under-recorded conditions. While the method is currently a research tool, further development could enable health systems to better estimate under-recorded mental health conditions and identify records that may require closer review. This could lead to more comprehensive care for veterans and potentially be applied to other health systems facing similar challenges.
Beyond the Headlines
The study raises important questions about the visibility of mental health conditions in medical records and the role of AI in addressing these challenges. The use of machine learning to identify under-recorded conditions could transform how health systems approach mental health care, making it more data-driven and comprehensive. This approach could also be applied to other conditions that are difficult to track, such as opioid use disorder and PTSD, potentially leading to broader improvements in healthcare delivery.
