Predicting Long-Term Outcome After Traumatic Brain Injury Using Repeated Measurements of Glasgow Coma Scale and Data Mining Methods

Hsueh Yi Lu, Tzu Chi Li, Yong Kwang Tu, Jui Chang Tsai, Hong Shiee Lai, Lu Ting Kuo

Research output: Contribution to journalArticle

11 Citations (Scopus)

Abstract

Previous studies have identified some clinical parameters for predicting long-term functional recovery and mortality after traumatic brain injury (TBI). Here, data mining methods were combined with serial Glasgow Coma Scale (GCS) scores and clinical and laboratory parameters to predict 6-month functional outcome and mortality in patients with TBI. Data of consecutive adult patients presenting at a trauma center with moderate-to-severe head injury were retrospectively analyzed. Clinical parameters including serial GCS measurements at emergency department, 7th day, and 14th day and laboratory data were included for analysis (n = 115). We employed artificial neural network (ANN), naïve Bayes (NB), decision tree, and logistic regression to predict mortality and functional outcomes at 6 months after TBI. Favorable functional outcome was achieved by 34.8 % of the patients, and overall 6-month mortality was 25.2 %. For 6-month functional outcome prediction, ANN was the best model, with an area under the receiver operating characteristic curve (AUC) of 96.13 %, sensitivity of 83.50 %, and specificity of 89.73 %. The best predictive model for mortality was NB with AUC of 91.14 %, sensitivity of 81.17 %, and specificity of 90.65 %. Sensitivity analysis demonstrated GCS measurements on the 7th and 14th day and difference between emergency room and 14th day GCS score as the most influential attributes both in mortality and functional outcome prediction models. Analysis of serial GCS measurements using data mining methods provided additional predictive information in relation to 6-month mortality and functional outcome in patients with moderate-to-severe TBI.

Original languageEnglish
Article number14
JournalJournal of Medical Systems
Volume39
Issue number2
DOIs
Publication statusPublished - Jan 31 2015
Externally publishedYes

Fingerprint

Glasgow Coma Scale
Data Mining
Data mining
Brain
Mortality
Emergency rooms
Neural networks
Area Under Curve
Decision trees
Hospital Emergency Service
Sensitivity analysis
Logistics
Sensitivity and Specificity
Decision Trees
Trauma Centers
Recovery
Traumatic Brain Injury
Craniocerebral Trauma
ROC Curve
Logistic Models

Keywords

  • Data mining
  • Glasgow coma scale
  • Mortality
  • Projections and predictions
  • Traumatic brain injury

ASJC Scopus subject areas

  • Medicine (miscellaneous)
  • Information Systems
  • Health Informatics
  • Health Information Management

Cite this

Predicting Long-Term Outcome After Traumatic Brain Injury Using Repeated Measurements of Glasgow Coma Scale and Data Mining Methods. / Lu, Hsueh Yi; Li, Tzu Chi; Tu, Yong Kwang; Tsai, Jui Chang; Lai, Hong Shiee; Kuo, Lu Ting.

In: Journal of Medical Systems, Vol. 39, No. 2, 14, 31.01.2015.

Research output: Contribution to journalArticle

@article{908f7980bef54d99b912db8c338ac130,
title = "Predicting Long-Term Outcome After Traumatic Brain Injury Using Repeated Measurements of Glasgow Coma Scale and Data Mining Methods",
abstract = "Previous studies have identified some clinical parameters for predicting long-term functional recovery and mortality after traumatic brain injury (TBI). Here, data mining methods were combined with serial Glasgow Coma Scale (GCS) scores and clinical and laboratory parameters to predict 6-month functional outcome and mortality in patients with TBI. Data of consecutive adult patients presenting at a trauma center with moderate-to-severe head injury were retrospectively analyzed. Clinical parameters including serial GCS measurements at emergency department, 7th day, and 14th day and laboratory data were included for analysis (n = 115). We employed artificial neural network (ANN), na{\"i}ve Bayes (NB), decision tree, and logistic regression to predict mortality and functional outcomes at 6 months after TBI. Favorable functional outcome was achieved by 34.8 {\%} of the patients, and overall 6-month mortality was 25.2 {\%}. For 6-month functional outcome prediction, ANN was the best model, with an area under the receiver operating characteristic curve (AUC) of 96.13 {\%}, sensitivity of 83.50 {\%}, and specificity of 89.73 {\%}. The best predictive model for mortality was NB with AUC of 91.14 {\%}, sensitivity of 81.17 {\%}, and specificity of 90.65 {\%}. Sensitivity analysis demonstrated GCS measurements on the 7th and 14th day and difference between emergency room and 14th day GCS score as the most influential attributes both in mortality and functional outcome prediction models. Analysis of serial GCS measurements using data mining methods provided additional predictive information in relation to 6-month mortality and functional outcome in patients with moderate-to-severe TBI.",
keywords = "Data mining, Glasgow coma scale, Mortality, Projections and predictions, Traumatic brain injury",
author = "Lu, {Hsueh Yi} and Li, {Tzu Chi} and Tu, {Yong Kwang} and Tsai, {Jui Chang} and Lai, {Hong Shiee} and Kuo, {Lu Ting}",
year = "2015",
month = "1",
day = "31",
doi = "10.1007/s10916-014-0187-x",
language = "English",
volume = "39",
journal = "Journal of Medical Systems",
issn = "0148-5598",
publisher = "Springer New York",
number = "2",

}

TY - JOUR

T1 - Predicting Long-Term Outcome After Traumatic Brain Injury Using Repeated Measurements of Glasgow Coma Scale and Data Mining Methods

AU - Lu, Hsueh Yi

AU - Li, Tzu Chi

AU - Tu, Yong Kwang

AU - Tsai, Jui Chang

AU - Lai, Hong Shiee

AU - Kuo, Lu Ting

PY - 2015/1/31

Y1 - 2015/1/31

N2 - Previous studies have identified some clinical parameters for predicting long-term functional recovery and mortality after traumatic brain injury (TBI). Here, data mining methods were combined with serial Glasgow Coma Scale (GCS) scores and clinical and laboratory parameters to predict 6-month functional outcome and mortality in patients with TBI. Data of consecutive adult patients presenting at a trauma center with moderate-to-severe head injury were retrospectively analyzed. Clinical parameters including serial GCS measurements at emergency department, 7th day, and 14th day and laboratory data were included for analysis (n = 115). We employed artificial neural network (ANN), naïve Bayes (NB), decision tree, and logistic regression to predict mortality and functional outcomes at 6 months after TBI. Favorable functional outcome was achieved by 34.8 % of the patients, and overall 6-month mortality was 25.2 %. For 6-month functional outcome prediction, ANN was the best model, with an area under the receiver operating characteristic curve (AUC) of 96.13 %, sensitivity of 83.50 %, and specificity of 89.73 %. The best predictive model for mortality was NB with AUC of 91.14 %, sensitivity of 81.17 %, and specificity of 90.65 %. Sensitivity analysis demonstrated GCS measurements on the 7th and 14th day and difference between emergency room and 14th day GCS score as the most influential attributes both in mortality and functional outcome prediction models. Analysis of serial GCS measurements using data mining methods provided additional predictive information in relation to 6-month mortality and functional outcome in patients with moderate-to-severe TBI.

AB - Previous studies have identified some clinical parameters for predicting long-term functional recovery and mortality after traumatic brain injury (TBI). Here, data mining methods were combined with serial Glasgow Coma Scale (GCS) scores and clinical and laboratory parameters to predict 6-month functional outcome and mortality in patients with TBI. Data of consecutive adult patients presenting at a trauma center with moderate-to-severe head injury were retrospectively analyzed. Clinical parameters including serial GCS measurements at emergency department, 7th day, and 14th day and laboratory data were included for analysis (n = 115). We employed artificial neural network (ANN), naïve Bayes (NB), decision tree, and logistic regression to predict mortality and functional outcomes at 6 months after TBI. Favorable functional outcome was achieved by 34.8 % of the patients, and overall 6-month mortality was 25.2 %. For 6-month functional outcome prediction, ANN was the best model, with an area under the receiver operating characteristic curve (AUC) of 96.13 %, sensitivity of 83.50 %, and specificity of 89.73 %. The best predictive model for mortality was NB with AUC of 91.14 %, sensitivity of 81.17 %, and specificity of 90.65 %. Sensitivity analysis demonstrated GCS measurements on the 7th and 14th day and difference between emergency room and 14th day GCS score as the most influential attributes both in mortality and functional outcome prediction models. Analysis of serial GCS measurements using data mining methods provided additional predictive information in relation to 6-month mortality and functional outcome in patients with moderate-to-severe TBI.

KW - Data mining

KW - Glasgow coma scale

KW - Mortality

KW - Projections and predictions

KW - Traumatic brain injury

UR - http://www.scopus.com/inward/record.url?scp=84925276775&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84925276775&partnerID=8YFLogxK

U2 - 10.1007/s10916-014-0187-x

DO - 10.1007/s10916-014-0187-x

M3 - Article

VL - 39

JO - Journal of Medical Systems

JF - Journal of Medical Systems

SN - 0148-5598

IS - 2

M1 - 14

ER -