Evaluating classification power of linked admission data sources with text mining

Simon Kocbek; Lawrence Cavedon; David Martinez; Christopher Ashley Bain; Chris Mac Manus; Gholamreza Haffari; Ingrid Zukerman; Karin Verspoor

Evaluating classification power of linked admission data sources with text mining

Simon Kocbek, Lawrence Cavedon, David Martinez, Christopher Ashley Bain, Chris Mac Manus, Gholamreza Haffari, Ingrid Zukerman, Karin Verspoor

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

Abstract

Lung cancer is a leading cause of death in developed countries. The paper presents a text mining system using Support Vector Machines for detecting lung cancer admissions. Performance of the system using different clinical data sources is evaluated. We use radiology reports as an initial data source and add other sources, such as pathology reports and patient demographics. Results show that linking data sources significantly improves classification performance with a maximum F-Score improvement of 0.057.

Original language	English
Title of host publication	Proceedings of the Scientific Stream at Big Data in Health Analytics 2015 (BigData 2015)
Editors	Katrina Barbuto, Louise Schaper, Karin Verspoor
Place of Publication	Aachen Germany
Publisher	Ruzica Piskac
Pages	1 - 7
Number of pages	7
Volume	1468
Publication status	Published - 2015
Event	Big Data in Health Analytics 2015 - Swissotel Sydney, Sydney, Australia Duration: 20 Oct 2015 → 21 Oct 2015 http://ceur-ws.org/Vol-1468/ (CEUR Workshop Proceedings) https://www.hisa.org.au/bigdata/science2015/

Conference

Conference	Big Data in Health Analytics 2015
Abbreviated title	BigData 2015
Country/Territory	Australia
City	Sydney
Period	20/10/15 → 21/10/15
Other	We are pleased to invite you to respond to our Call for Submissions for Big Data 2015, Australasia’s Big Data In Biomedicine and Healthcare Conference, to be held from 20-21 October 2015 at the Swissotel Sydney. With a strong focus on collaboration, this year’s theme – Big data analytics: Leveraging capability in healthcare – reflects the opportunity for healthcare practitioners, information specialists in healthcare environments, university and research institute scientists to share their knowledge.
Internet address	http://ceur-ws.org/Vol-1468/ (CEUR Workshop Proceedings) https://www.hisa.org.au/bigdata/science2015/

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Cite this

Kocbek, S., Cavedon, L., Martinez, D., Bain, C. A., Mac Manus, C., Haffari, G., Zukerman, I., & Verspoor, K. (2015). Evaluating classification power of linked admission data sources with text mining. In K. Barbuto, L. Schaper, & K. Verspoor (Eds.), Proceedings of the Scientific Stream at Big Data in Health Analytics 2015 (BigData 2015) (Vol. 1468, pp. 1 - 7). Ruzica Piskac.

@inproceedings{4aab1d72702549adb0ceb8b23175e5fd,

title = "Evaluating classification power of linked admission data sources with text mining",

abstract = "Lung cancer is a leading cause of death in developed countries. The paper presents a text mining system using Support Vector Machines for detecting lung cancer admissions. Performance of the system using different clinical data sources is evaluated. We use radiology reports as an initial data source and add other sources, such as pathology reports and patient demographics. Results show that linking data sources significantly improves classification performance with a maximum F-Score improvement of 0.057.",

author = "Simon Kocbek and Lawrence Cavedon and David Martinez and Bain, {Christopher Ashley} and {Mac Manus}, Chris and Gholamreza Haffari and Ingrid Zukerman and Karin Verspoor",

year = "2015",

language = "English",

volume = "1468",

pages = "1 -- 7",

editor = "Katrina Barbuto and Louise Schaper and Karin Verspoor",

booktitle = "Proceedings of the Scientific Stream at Big Data in Health Analytics 2015 (BigData 2015)",

publisher = "Ruzica Piskac",

note = "Big Data in Health Analytics 2015, BigData 2015 ; Conference date: 20-10-2015 Through 21-10-2015",

url = "http://ceur-ws.org/Vol-1468/, https://www.hisa.org.au/bigdata/science2015/",

}

Kocbek, S, Cavedon, L, Martinez, D, Bain, CA, Mac Manus, C, Haffari, G , Zukerman, I & Verspoor, K 2015, Evaluating classification power of linked admission data sources with text mining. in K Barbuto, L Schaper & K Verspoor (eds), Proceedings of the Scientific Stream at Big Data in Health Analytics 2015 (BigData 2015). vol. 1468, Ruzica Piskac, Aachen Germany, pp. 1 - 7, Big Data in Health Analytics 2015, Sydney, New South Wales, Australia, 20/10/15.

Evaluating classification power of linked admission data sources with text mining. / Kocbek, Simon; Cavedon, Lawrence; Martinez, David et al.
Proceedings of the Scientific Stream at Big Data in Health Analytics 2015 (BigData 2015). ed. / Katrina Barbuto; Louise Schaper; Karin Verspoor. Vol. 1468 Aachen Germany: Ruzica Piskac, 2015. p. 1 - 7.

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

TY - GEN

T1 - Evaluating classification power of linked admission data sources with text mining

AU - Kocbek, Simon

AU - Cavedon, Lawrence

AU - Martinez, David

AU - Bain, Christopher Ashley

AU - Mac Manus, Chris

AU - Haffari, Gholamreza

AU - Zukerman, Ingrid

AU - Verspoor, Karin

PY - 2015

Y1 - 2015

N2 - Lung cancer is a leading cause of death in developed countries. The paper presents a text mining system using Support Vector Machines for detecting lung cancer admissions. Performance of the system using different clinical data sources is evaluated. We use radiology reports as an initial data source and add other sources, such as pathology reports and patient demographics. Results show that linking data sources significantly improves classification performance with a maximum F-Score improvement of 0.057.

AB - Lung cancer is a leading cause of death in developed countries. The paper presents a text mining system using Support Vector Machines for detecting lung cancer admissions. Performance of the system using different clinical data sources is evaluated. We use radiology reports as an initial data source and add other sources, such as pathology reports and patient demographics. Results show that linking data sources significantly improves classification performance with a maximum F-Score improvement of 0.057.

UR - http://ceur-ws.org/Vol-1468/bd2015_kocbek.pdf

M3 - Conference Paper

VL - 1468

SP - 1

EP - 7

BT - Proceedings of the Scientific Stream at Big Data in Health Analytics 2015 (BigData 2015)

A2 - Barbuto, Katrina

A2 - Schaper, Louise

A2 - Verspoor, Karin

PB - Ruzica Piskac

CY - Aachen Germany

T2 - Big Data in Health Analytics 2015

Y2 - 20 October 2015 through 21 October 2015

ER -

Evaluating classification power of linked admission data sources with text mining

Abstract

Conference

UN SDGs

Other files and links

Cite this