Veterans’ Hidden Struggles: Spotting Unseen Self‑Harm in Health Records
Health Records Miss Signs of Self‑Harm, Especially Among Veterans
Problem
Electronic health records often fail to flag self‑harm, especially in veteran populations. Because clinicians only mark clear cases, the data lacks true “negative” examples.Consequence
The missing negatives make it difficult to estimate how many people are actually at risk.Solution
Researchers applied Positive and Unlabeled learning, a technique that works well when only confirmed positives are labeled—as is common in medical databases.
- What It Does
- Estimates the true number of patients with self‑harm behaviors.
Identifies those who may be hiding these issues.
Findings
In a study of veteran electronic records, the method uncovered many veterans who had not been formally diagnosed but exhibited patterns linked to self‑harm.Implications
These hidden cases could benefit from early support and treatment. Current coding practices leave many at risk unnoticed.Next Steps
Treating data as “positive or unlabeled” allows clinicians to better identify those needing help, potentially improving preventive care and saving lives.
