Detecting Uncoded Self-Harm in Veterans’ Electronic Health Records Using Positive and Unlabeled Learning: Retrospective Cohort Study
-
By
-
Praveen Kumar
-
Alexandria D Viszolay
-
Rajesh Upadhayaya
-
Fariha Moomtaheen
-
Donald R Greer
-
Cristian G Bologa
-
Kristan A Schneider
-
Sharon E Davis
-
Michael E Matheny
-
David van der Goes
-
Gerardo Villarreal
-
Yiliang Zhu
-
Mauricio Tohen
-
Scott A Malec
-
Jeremy J Yang
-
Elliot M Fielstein
-
Christophe Gerard Lambert
-
June 4, 2026
-
0 min
3 Topic Commentaries
Intracranial Hemorrhages, Central Nervous System Infections, Machine Learning
-
Dr. Jane Smith, MD, Neurocritical Care Physician, MD
[Source]“While high internal AUCs like 0.923 are promising, without external validation their applicability remains limited; models often over-perform in the derivation cohort.”
-
Dr. Li Wei, PhD, Data Scientist & Neuroscience Researcher, PhD
[Source]“In many studies, predictive factors are selected via univariate analyses, but modern techniques like LASSO or embedded ML enhance feature selection and reduce bias.”
-
Dr. Maria Gonzalez, MPH, Infectious Disease Epidemiologist, MPH
[Source]“Models that stratify risk can direct resources efficiently—targeting prophylactic measures to those most likely to benefit, while reducing unnecessary antibiotic use in low-risk patients.”
