Skip to content

VINCI-AppliedNLP/PTSDChecklist

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

20 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

PTSDChecklist

NLP System to extract PCL scores from clinic notes. The PTSD Checklist (PCL) is a standardized self-report rating scale for PTSD comprising of several items that correspond to the key symptoms of PTSD. Three types of the PCL exist: 1) PCL-M (PCL-Military) is specific to PTSD caused by military experiences; 2) PCL-C (PCL-Civilian) is applied generally to any traumatic event; and 3) PCL-S (PCL-Specific) is applied to a specific stressful life situation. There are also two versions of PCL: 1) PCL-4 based on DSM-IV, which was in use prior to 2013 and 2) PCL-5 based on DSM-5, which came in effect in 2013. The goal of this project is to apply Natural Language Processing (NLP) techniques to extract mentions of the specific PCL administered to a patient and the total score that was acquired as a result.

Example values extracted:

“PCL score = 77.”

“PCL score was a 55.”

“PCL): 65”

“PCL-S score of 62”

“PCL=41”

“PCL-5): current score of 58”

“PCL was administered; score was 55.”

“PCL-M score = 75”

“PCL-S) yielded a score of 67%”

“PCL-S, getting a score of 39”

Validation

This system was validated for precision with 1000 random notes from Psycotherapy visits being selected.

It was validated to reach 98% positive predictive value. We did not measure sensitivity. We found PCL values in approximately 65% of these documents.

  • REMARK: BFG removed ORD in the history An IE system for the PTSD Checklist Score(PCL)

Analysis

PTSDEBP-RTH This project set out to determine whether natural language processing (NLP) extraction of PCL scores from TIU clinical notes could meaningfully supplement the structured CDW data in a subset of patients receiving EBP between January 2017 and September 2024.

About

An IE system for the PTSD Checklist Score(PCL)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors