Temporal abstraction for feature extraction: A comparative case study in prediction from intensive care monitoring data

Marion Verduijn, Lucia Sacchi, Niels Peek, Riccardo Bellazzi, Evert de Jonge, Bas A J M de Mol

Research output: Contribution to journalArticle

30 Citations (Scopus)

Abstract

Objectives: To compare two temporal abstraction procedures for the extraction of meta features from monitoring data. Feature extraction prior to predictive modeling is a common strategy in prediction from temporal data. A fundamental dilemma in this strategy, however, is the extent to which the extraction should be guided by domain knowledge, and to which extent it should be guided by the available data. The two temporal abstraction procedures compared in this case study differ in this respect. Methods and material: The first temporal abstraction procedure derives symbolic descriptions from the data that are predefined using existing concepts from the medical language. In the second procedure, a large space of numerical meta features is searched through to discover relevant features from the data. These procedures were applied to a prediction problem from intensive care monitoring data. The predictive value of the resulting meta features were compared, and based on each type of features, a class probability tree model was developed. Results: The numerical meta features extracted by the second procedure were found to be more informative than the symbolic meta features of the first procedure in the case study, and a superior predictive performance was observed for the associated tree model. Conclusion: The findings indicate that for prediction from monitoring data, induction of numerical meta features from data is preferable to extraction of symbolic meta features using existing clinical concepts.

Original languageEnglish
Pages (from-to)1-12
Number of pages12
JournalArtificial Intelligence in Medicine
Volume41
Issue number1
DOIs
Publication statusPublished - Sep 2007

Fingerprint

Critical Care
Feature extraction
Monitoring
Language

Keywords

  • Cardiac surgery
  • Feature extraction
  • Intensive care
  • Monitoring data
  • Prognosis
  • Temporal abstraction
  • Temporal classification

ASJC Scopus subject areas

  • Artificial Intelligence
  • Medicine(all)

Cite this

Temporal abstraction for feature extraction : A comparative case study in prediction from intensive care monitoring data. / Verduijn, Marion; Sacchi, Lucia; Peek, Niels; Bellazzi, Riccardo; de Jonge, Evert; de Mol, Bas A J M.

In: Artificial Intelligence in Medicine, Vol. 41, No. 1, 09.2007, p. 1-12.

Research output: Contribution to journalArticle

Verduijn, Marion ; Sacchi, Lucia ; Peek, Niels ; Bellazzi, Riccardo ; de Jonge, Evert ; de Mol, Bas A J M. / Temporal abstraction for feature extraction : A comparative case study in prediction from intensive care monitoring data. In: Artificial Intelligence in Medicine. 2007 ; Vol. 41, No. 1. pp. 1-12.
@article{2412c903c04749f0aa968420ecd18064,
title = "Temporal abstraction for feature extraction: A comparative case study in prediction from intensive care monitoring data",
abstract = "Objectives: To compare two temporal abstraction procedures for the extraction of meta features from monitoring data. Feature extraction prior to predictive modeling is a common strategy in prediction from temporal data. A fundamental dilemma in this strategy, however, is the extent to which the extraction should be guided by domain knowledge, and to which extent it should be guided by the available data. The two temporal abstraction procedures compared in this case study differ in this respect. Methods and material: The first temporal abstraction procedure derives symbolic descriptions from the data that are predefined using existing concepts from the medical language. In the second procedure, a large space of numerical meta features is searched through to discover relevant features from the data. These procedures were applied to a prediction problem from intensive care monitoring data. The predictive value of the resulting meta features were compared, and based on each type of features, a class probability tree model was developed. Results: The numerical meta features extracted by the second procedure were found to be more informative than the symbolic meta features of the first procedure in the case study, and a superior predictive performance was observed for the associated tree model. Conclusion: The findings indicate that for prediction from monitoring data, induction of numerical meta features from data is preferable to extraction of symbolic meta features using existing clinical concepts.",
keywords = "Cardiac surgery, Feature extraction, Intensive care, Monitoring data, Prognosis, Temporal abstraction, Temporal classification",
author = "Marion Verduijn and Lucia Sacchi and Niels Peek and Riccardo Bellazzi and {de Jonge}, Evert and {de Mol}, {Bas A J M}",
year = "2007",
month = "9",
doi = "10.1016/j.artmed.2007.06.003",
language = "English",
volume = "41",
pages = "1--12",
journal = "Artificial Intelligence in Medicine",
issn = "0933-3657",
publisher = "Elsevier",
number = "1",

}

TY - JOUR

T1 - Temporal abstraction for feature extraction

T2 - A comparative case study in prediction from intensive care monitoring data

AU - Verduijn, Marion

AU - Sacchi, Lucia

AU - Peek, Niels

AU - Bellazzi, Riccardo

AU - de Jonge, Evert

AU - de Mol, Bas A J M

PY - 2007/9

Y1 - 2007/9

N2 - Objectives: To compare two temporal abstraction procedures for the extraction of meta features from monitoring data. Feature extraction prior to predictive modeling is a common strategy in prediction from temporal data. A fundamental dilemma in this strategy, however, is the extent to which the extraction should be guided by domain knowledge, and to which extent it should be guided by the available data. The two temporal abstraction procedures compared in this case study differ in this respect. Methods and material: The first temporal abstraction procedure derives symbolic descriptions from the data that are predefined using existing concepts from the medical language. In the second procedure, a large space of numerical meta features is searched through to discover relevant features from the data. These procedures were applied to a prediction problem from intensive care monitoring data. The predictive value of the resulting meta features were compared, and based on each type of features, a class probability tree model was developed. Results: The numerical meta features extracted by the second procedure were found to be more informative than the symbolic meta features of the first procedure in the case study, and a superior predictive performance was observed for the associated tree model. Conclusion: The findings indicate that for prediction from monitoring data, induction of numerical meta features from data is preferable to extraction of symbolic meta features using existing clinical concepts.

AB - Objectives: To compare two temporal abstraction procedures for the extraction of meta features from monitoring data. Feature extraction prior to predictive modeling is a common strategy in prediction from temporal data. A fundamental dilemma in this strategy, however, is the extent to which the extraction should be guided by domain knowledge, and to which extent it should be guided by the available data. The two temporal abstraction procedures compared in this case study differ in this respect. Methods and material: The first temporal abstraction procedure derives symbolic descriptions from the data that are predefined using existing concepts from the medical language. In the second procedure, a large space of numerical meta features is searched through to discover relevant features from the data. These procedures were applied to a prediction problem from intensive care monitoring data. The predictive value of the resulting meta features were compared, and based on each type of features, a class probability tree model was developed. Results: The numerical meta features extracted by the second procedure were found to be more informative than the symbolic meta features of the first procedure in the case study, and a superior predictive performance was observed for the associated tree model. Conclusion: The findings indicate that for prediction from monitoring data, induction of numerical meta features from data is preferable to extraction of symbolic meta features using existing clinical concepts.

KW - Cardiac surgery

KW - Feature extraction

KW - Intensive care

KW - Monitoring data

KW - Prognosis

KW - Temporal abstraction

KW - Temporal classification

UR - http://www.scopus.com/inward/record.url?scp=34548277939&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=34548277939&partnerID=8YFLogxK

U2 - 10.1016/j.artmed.2007.06.003

DO - 10.1016/j.artmed.2007.06.003

M3 - Article

C2 - 17698331

AN - SCOPUS:34548277939

VL - 41

SP - 1

EP - 12

JO - Artificial Intelligence in Medicine

JF - Artificial Intelligence in Medicine

SN - 0933-3657

IS - 1

ER -