BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20241120T082410Z
LOCATION:HG E Haupthalle
DTSTART;TZID=Europe/Stockholm:20240604T193000
DTEND;TZID=Europe/Stockholm:20240604T213000
UID:submissions.pasc-conference.org_PASC24_sess151_pos147@linklings.com
SUMMARY:P45 - Scaled Life Event Extraction using High Performance Computin
 g for Acute Veteran Suicide Risk Prediction
DESCRIPTION:Poster\n\nDestinee Morrow, Rafael Zamora-Resendiz, and Mahamad
  Mahmoud (Lawrence Berkeley National Laboratory); Jean Beckham and Nathan 
 Kimbrel (VA Durham Health Care); Benjamin McMahon (Los Alamos National Lab
 oratory); and Silvia Crivelli (Lawrence Berkeley National Laboratory)\n\nP
 redictive models of suicide risk have focused on predictors extracted from
  structured data found in electronic health records (EHR), with limited co
 nsideration of negative life events (LE) expressed in unstructured clinica
 l text such as housing instability, marital troubles, etc. Additionally, t
 here has been limited work in large-scale analysis of natural language pro
 cessing (NLP) derived predictors for suicide risk and integration of extra
 cted LE into longitudinal and predictive models of suicide risk. Our study
  aims to expand upon previous research, showing how large language models 
 (LLM) and high-performance computing (HPC) can be used to annotate LE span
 ning over 22 years in the Veterans Affairs (VA) corporate data warehouse (
 CDW) with enriched sensitivity and demonstrate trends for acute suicide ri
 sk. Many Veteran timelines reference more than one LE in unstructured clin
 ical text by the time a suicide-related diagnosis was recorded. Longitudin
 al data from extractions serve as acute predictors of suicide-related even
 ts. Preliminary analysis of ascertain administrative bias in NLP extractio
 ns show many mentions occur prior to triaging by case-coordinators. Lastly
 , LE provide essential input that improves the performance of predictive m
 odeling concerning suicide-related events.
END:VEVENT
END:VCALENDAR