Preprocessing of natural language process variables using a data-driven method improves the association with suicide risk in a large veterans affairs population

Siting Li, Maxwell Levis, Monica Dimambro, Weiyi Wu, Joshua Levy, Brian Shiner, Jiang Gui

Abstract: Highlights•AMC preprocessing improves associations between NLP variables and suicide risk.•Over 90 % of AMC-processed NLP variables are significantly associated with suicide.•AMC outperforms quantile categorization in whole and undersampled cohorts.•AMC refines risk modeling for suicide prevention in clinical settings.•AMC may enhance NLP-based suicide risk prediction in Veterans Affairs EHR notes.