Machine Learning Method Reveals Hidden Self-Harm Histories in Veterans' Medical Records, Quadrupling Detection Rates
核心洞察
A novel machine learning method called PULSNAR estimated that 7.9% of VHA patients had documented self-harm (搜索) history, more than four times the 1.85% captured by diagnosis codes alone.
Diagnosis codes captured only about one-fourth of clinically documented self-harm (搜索) history in a study analyzing electronic health records from over 1.3 million veterans.
Even when self-harm (搜索) appeared in diagnosis codes, it was missing from problem lists in 77.4% of cases, revealing a critical visibility gap in clinical summaries.
A study led by researchers at The University of New Mexico School of Medicine (搜索) has uncovered a substantial gap in how health systems track self-harm (搜索) history, finding that standard diagnosis codes capture only about one-fourth of clinically documented cases. The research, published in the Journal of Medical Internet Research, analyzed electronic health records for more than 1.3 million patients served by the Veterans Health Administration (VHA) and demonstrated that a novel machine learning approach can surface self-harm histories that remain buried in unstructured clinical notes.
The findings carry significant implications for mental health service planning, clinical research, and ultimately patient care. Past self-harm (搜索) is one of the most important predictors of future self-harm and suicide risk, and it can shape how clinicians approach conditions such as depression (搜索), PTSD (搜索), bipolar disorder (搜索), substance use, and traumatic brain injury (搜索) that frequently co-occur with self-harm.
A Fourfold Detection Gap
Following expert chart review and statistical calibration, the researchers estimated that documented self-harm (搜索) was present in approximately 7.9% of patients seen by VHA clinicians — more than four times the 1.85% visible through diagnosis codes alone.
"For research and planning, if we only count what is easy to see in diagnosis codes, we may substantially underestimate the need for mental health services," said Christophe Lambert, PhD, professor and interim chief of the Division of Translational Informatics in the UNM School of Medicine's Department of Internal Medicine and the study's corresponding author. "Better measurement can help health systems plan better, help researchers study care more accurately and eventually help clinicians know when a patient may need a closer look."
Problem Lists: Another Visibility Gap
The study also revealed deficiencies in problem lists — the notations providers compile of their patients' health conditions, meant to flag important conditions for clinical teams. Among veterans with a diagnosis code for self-harm (搜索), only 22.6% had self-harm or a history of self-harm listed on their VHA problem list. This means that even when self-harm appeared in diagnosis codes, it was often missing from one of the record's most visible summary fields.
"This is a systems-level visibility problem," Lambert said. "The record can be enormous. In our chart review, some patient records had more than 500,000 lines of notes. No clinician can be expected to read all of that during a normal visit."
The PULSNAR Methodology
The research team employed a method called PULSNAR — Positive Unlabeled Learning Selected Not At Random — specifically designed for the complexities of real-world health data. Most machine learning methods require clear examples of both "yes" and "no" cases, but in medical records, a missing diagnosis code does not prove that a patient never had the condition.
PULSNAR works within that uncertainty. It learns from patients who do have a diagnosis code, then estimates how many similar patients might be present among those without a code. Its key advantage is that it does not assume coded cases are random and accounts for the fact that some cases are more likely to be coded than others.
"Medical records can make self-harm (搜索) hard to see in more than one way," said Praveen Kumar, PhD, the study's first author. "Sometimes the history is in a clinician's note but not in the diagnosis codes. Other times, the record may contain risk factors, injuries, poisonings, or behaviors that are consistent with self-harm, even though the record alone does not prove what happened or why."
Kumar noted that the method can help flag both patterns for review. The study verified the first pattern — where evidence was already present in clinical notes — but confirming the second pattern, involving risk factors and behaviors, would require talking with patients or using information beyond the medical record.
A Broader Research Program
The self-harm (搜索) study is part of a broader research program using positive-and-unlabeled learning to find conditions that may be under-recorded in standard medical data. The team has already published a related study using this approach to detect under-coded opioid use disorder (搜索), and ongoing work is extending it to other conditions where the medical record may not show the full picture, including unrecognized PTSD (搜索), depression (搜索), bipolar disorder (搜索), and sleep disorders.
The research team brought together expertise from the UNM Health Sciences Center, the Raymond G. Murphy Veterans Affairs Medical Center, Vanderbilt University Medical Center, the VA Tennessee Valley Healthcare System, the VA Office of Mental Health, Greer Black Company, and the UNM Department of Economics, spanning medical informatics, computer science, psychiatry, biomedical informatics, economics, statistics, and health services research.
The authors emphasized that VHA already uses specialized suicide and overdose reporting tools and does not rely only on diagnosis codes or problem lists to monitor suicide risk. This study addressed a different but related question: how much past self-harm (搜索) history is visible in the parts of the record that researchers, care teams, and health systems can most easily quantify and review at scale.
The investigators stressed that the method remains a research tool and is not ready to be used by itself in clinical care. However, with further development, it could help health systems better estimate under-recorded mental health conditions, find documented history that is not clearly visible, and identify records that may warrant closer review.
"Self-harm (搜索) history matters too much to stay buried in records that are not practical to review line by line during routine care," Lambert said. "Our work is about helping researchers and health systems find documented history and clinically relevant patterns in the data, so care teams can have a more complete picture of the people they serve."