A feature representation learning method for temporal datasets

Ward Van Breda, Mark Hoogendoorn, A E Eiben, Gerhard Andersson, Heleen Riper, Jeroen Ruwaard, Kristofer Vernmark

Research output: Contribution to conferencePaperAcademic

Abstract

—Predictive modeling of future health states can greatly contribute to more effective health care. Healthcare professionals can for example act in a more proactive way or predictions can drive more automated ways of therapy. However, the task is very challenging. Future developments likely depend on observations in the (recent) past, but how can we capture this history in features to generate accurate predictive models? And what length of history should we consider? We propose a framework that is able to generate patient tailored features from observations of the recent history that maximize predictive performance. For a case study in the domain of depression we find that using this method new data representations can be generated that increase the predictive performance significantly.
Original languageEnglish
Number of pages8
DOIs
Publication statusPublished - 2016

Cite this

Van Breda, W., Hoogendoorn, M., Eiben, A. E., Andersson, G., Riper, H., Ruwaard, J., & Vernmark, K. (2016). A feature representation learning method for temporal datasets. https://doi.org/10.1109/SSCI.2016.7849890
Van Breda, Ward ; Hoogendoorn, Mark ; Eiben, A E ; Andersson, Gerhard ; Riper, Heleen ; Ruwaard, Jeroen ; Vernmark, Kristofer. / A feature representation learning method for temporal datasets. 8 p.
@conference{b62d4d47626f40368a30577eea64a3ce,
title = "A feature representation learning method for temporal datasets",
abstract = "—Predictive modeling of future health states can greatly contribute to more effective health care. Healthcare professionals can for example act in a more proactive way or predictions can drive more automated ways of therapy. However, the task is very challenging. Future developments likely depend on observations in the (recent) past, but how can we capture this history in features to generate accurate predictive models? And what length of history should we consider? We propose a framework that is able to generate patient tailored features from observations of the recent history that maximize predictive performance. For a case study in the domain of depression we find that using this method new data representations can be generated that increase the predictive performance significantly.",
author = "{Van Breda}, Ward and Mark Hoogendoorn and Eiben, {A E} and Gerhard Andersson and Heleen Riper and Jeroen Ruwaard and Kristofer Vernmark",
year = "2016",
doi = "10.1109/SSCI.2016.7849890",
language = "English",

}

A feature representation learning method for temporal datasets. / Van Breda, Ward; Hoogendoorn, Mark; Eiben, A E; Andersson, Gerhard; Riper, Heleen; Ruwaard, Jeroen; Vernmark, Kristofer.

2016.

Research output: Contribution to conferencePaperAcademic

TY - CONF

T1 - A feature representation learning method for temporal datasets

AU - Van Breda, Ward

AU - Hoogendoorn, Mark

AU - Eiben, A E

AU - Andersson, Gerhard

AU - Riper, Heleen

AU - Ruwaard, Jeroen

AU - Vernmark, Kristofer

PY - 2016

Y1 - 2016

N2 - —Predictive modeling of future health states can greatly contribute to more effective health care. Healthcare professionals can for example act in a more proactive way or predictions can drive more automated ways of therapy. However, the task is very challenging. Future developments likely depend on observations in the (recent) past, but how can we capture this history in features to generate accurate predictive models? And what length of history should we consider? We propose a framework that is able to generate patient tailored features from observations of the recent history that maximize predictive performance. For a case study in the domain of depression we find that using this method new data representations can be generated that increase the predictive performance significantly.

AB - —Predictive modeling of future health states can greatly contribute to more effective health care. Healthcare professionals can for example act in a more proactive way or predictions can drive more automated ways of therapy. However, the task is very challenging. Future developments likely depend on observations in the (recent) past, but how can we capture this history in features to generate accurate predictive models? And what length of history should we consider? We propose a framework that is able to generate patient tailored features from observations of the recent history that maximize predictive performance. For a case study in the domain of depression we find that using this method new data representations can be generated that increase the predictive performance significantly.

U2 - 10.1109/SSCI.2016.7849890

DO - 10.1109/SSCI.2016.7849890

M3 - Paper

ER -