Language Annotation Modeling and Processing Toolkit CLAMP Hua Xu School of Biomedical Informatics University of Texas Health Science Center at Houston 1 The Transportability Problem of ID: 624919
Download Presentation The PPT/PDF document "Clinical" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Clinical Language Annotation, Modeling, and Processing Toolkit (CLAMP)
Hua XuSchool of Biomedical Informatics, University of Texas Health Science Center at Houston
1Slide2
The Transportability Problem of using clinical NLP systemsFrom one type of clinical notes to
anotherFrom one institute to another From one application to anotherNeed a solution for non-NLP experts to efficiently build high-performance NLP modules for individual applications!2Slide3
What is CLAMP?An IDE (integrated development environment) for building customized clinical NLP pipelines via GUIs
Annotating/analyzing clinical textTraining of ML-based modulesSpecifying rulesA general purpose clinical NLP system built on proven methods3NLP TasksRanking
Named entity recognition2009 i2b2, medication#22010
i2b2 problem, treatment, test
#2
2013
SHARe
/CLEF abbreviation
#1
UMLS encoding
2014 SemEval, disorder#1Relation extraction2012 i2b2 Temporal#12015 SemEval Disease-modifier#12015 BioCREATIVE Chemical-induced disease #1Slide4
CLAMP Demo 1 – Build a rule-based system to extract smoking status from clinical text
Input: sentences containing patient smoking informationOutput: three types of status for each smoking mention:Current Smoker: She is continuing to smokePast Smoker: She has a prior history of smoking although not currentlyNon-Smoker: She denies any tobacco use , alcohol use 4Slide5
CLAMP Demo 2 - Build a hybrid (machine learning + rules) system for extracting labtest concepts and values from clinical text
Input: discharge summariesOutput: lab test concepts mentioned in the text with attributes of:OffsetsNegationUMLS CUIsValue5Slide6
Availability CLAMP is available in two versions:CLAMP CMD (free)
CLAMP GUI (depends on the license)https://sbmi.uth.edu/ccb/resources/clamp.htmIt is not an open source software, but source codes are available for collaborators with appropriate licenses. We are looking for collaborators to co-develop the system! If interested, please contact: Hua.Xu@uth.tmc.edu Slide7
Acknowledgement
GrantsCPRIT R1307 NIGMS R01 GM102282NLM R01 LM010681
CollaboratorsHongfang Liu, PhDSerguei Pakhomov, PhDJason Hou, MD
Team members:
Jingqi
Wang
Min Jiang
Ergin
Soysal
Sungrim MoonJun XuYaoyun ZhangAnupama GururajYonghui WuNina SlimiKyle NguyenTolulola DawoduYukun ChenQiang Wei7Slide8
Thank you!
Questions?hua.xu@uth.tmc.edu
8