|
Title:
|
EXTRACTING FRAILTY STATUS FOR POST SURGICAL MORTALITY PREDICTION |
|
Author(s):
|
Qing Zeng-Treitler, Yijun Shao, Yan Cheng, Kristina Doing-Harris, Rashmee U. Shah, Charlene R. Weir and Bruce E. Bray |
|
ISBN:
|
978-989-8533-77-7 |
|
Editors:
|
Mário Macedo and Piet Kommers |
|
Year:
|
2018 |
|
Edition:
|
Single |
|
Keywords:
|
Frailty, Natural Language Processing, Predictive Modeling |
|
Type:
|
Full Paper |
|
First Page:
|
20 |
|
Last Page:
|
28 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
Frailty is increasingly recognized as a leading indicator of worsening health outcomes, even death. Frailty is also used implicitly by many clinicians for decision-making, but rarely in a systematic way. We developed a novel ontology-based natural language processing approach to extract frailty status from clinical notes and support retrospective clinical studies and prospective clinical decision making. The natural language processing classification of frailty description achieved an accuracy of 80.3%. We used the extracted frailty information along with other clinical variables to predict mortality after major cardiovascular procedures. The area under the curve for mortality prediction was 78.3% on the test data for a deep neural network model, compared to 74.9% using a support vector machine. |
|
|
|
|
|
|
