Tree-structured Subgroup Analysis of Receiver Operating Characteristic Curves for Diagnostic Tests

Caixia Li, Claus C. Glüer, Richard Eastell, Dieter Felsenberg, David M. Reid, Christian Roux, Ying Lu*

*Corresponding author for this work

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

Rationale and Objectives: Multiple diagnostic tests are often available for a disease. Their diagnostic accuracy may depend on the characteristics of testing subjects. The investigators propose a new tree-structured data-mining method that identifies subgroups and their corresponding diagnostic tests to achieve the maximum area under the receiver-operating characteristic curve. Materials and Methods: The Osteoporosis and Ultrasound Study is a prospectively designed, population-based European multicenter observational study to evaluate state-of-the-art diagnostic methods for assessing osteoporosis. A total 2837 women underwent dual x-ray absorptiometry (DXA) and quantitative ultrasound (QUS). Prevalent vertebral fractures were determined by a centralized radiology laboratory on the basis of radiographs. The data-mining algorithm includes three steps: defining the criteria for node splitting and selection of the best diagnostic test on the basis of the area under the curve, using a random forest to estimate the probability of DXA being the preferred diagnostic method for each participant, and building a single regression tree to describe subgroups for which either DXA or QUS is the more accurate test or for which the two tests are equivalent. Results: For participants with weights ≤54.5 kg, QUS had a higher area under the curve in identifying prevalent vertebral fracture. For participants whose weights were >58.5 kg and whose heights were ≤167.5 cm, DXA was better, and for the remaining participants, DXA and QUS had comparable accuracy and could be used interchangeably. Conclusions: The proposed tree-structured subgroup analysis successfully defines subgroups and their best diagnostic tests. The method can be used to develop optimal diagnostic strategies in personalized medicine.

Original languageEnglish
Pages (from-to)1529-1536
Number of pages8
JournalAcademic Radiology
Volume19
Issue number12
Early online date31 Oct 2012
DOIs
Publication statusPublished - 1 Dec 2012

Fingerprint

Routine Diagnostic Tests
ROC Curve
X-Rays
Data Mining
Osteoporosis
Area Under Curve
Weights and Measures
Precision Medicine
Radiology
Multicenter Studies
Observational Studies
Research Personnel
Population

Keywords

  • Classification and regression tree
  • Personalized medicine
  • Random forest
  • Receiver-operating characteristic curve
  • Subgroup analysis

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging

Cite this

Tree-structured Subgroup Analysis of Receiver Operating Characteristic Curves for Diagnostic Tests. / Li, Caixia; Glüer, Claus C.; Eastell, Richard; Felsenberg, Dieter; Reid, David M.; Roux, Christian; Lu, Ying.

In: Academic Radiology, Vol. 19, No. 12, 01.12.2012, p. 1529-1536.

Research output: Contribution to journalArticle

Li, Caixia ; Glüer, Claus C. ; Eastell, Richard ; Felsenberg, Dieter ; Reid, David M. ; Roux, Christian ; Lu, Ying. / Tree-structured Subgroup Analysis of Receiver Operating Characteristic Curves for Diagnostic Tests. In: Academic Radiology. 2012 ; Vol. 19, No. 12. pp. 1529-1536.
@article{532001ca709b405b90cafa712ecf1304,
title = "Tree-structured Subgroup Analysis of Receiver Operating Characteristic Curves for Diagnostic Tests",
abstract = "Rationale and Objectives: Multiple diagnostic tests are often available for a disease. Their diagnostic accuracy may depend on the characteristics of testing subjects. The investigators propose a new tree-structured data-mining method that identifies subgroups and their corresponding diagnostic tests to achieve the maximum area under the receiver-operating characteristic curve. Materials and Methods: The Osteoporosis and Ultrasound Study is a prospectively designed, population-based European multicenter observational study to evaluate state-of-the-art diagnostic methods for assessing osteoporosis. A total 2837 women underwent dual x-ray absorptiometry (DXA) and quantitative ultrasound (QUS). Prevalent vertebral fractures were determined by a centralized radiology laboratory on the basis of radiographs. The data-mining algorithm includes three steps: defining the criteria for node splitting and selection of the best diagnostic test on the basis of the area under the curve, using a random forest to estimate the probability of DXA being the preferred diagnostic method for each participant, and building a single regression tree to describe subgroups for which either DXA or QUS is the more accurate test or for which the two tests are equivalent. Results: For participants with weights ≤54.5 kg, QUS had a higher area under the curve in identifying prevalent vertebral fracture. For participants whose weights were >58.5 kg and whose heights were ≤167.5 cm, DXA was better, and for the remaining participants, DXA and QUS had comparable accuracy and could be used interchangeably. Conclusions: The proposed tree-structured subgroup analysis successfully defines subgroups and their best diagnostic tests. The method can be used to develop optimal diagnostic strategies in personalized medicine.",
keywords = "Classification and regression tree, Personalized medicine, Random forest, Receiver-operating characteristic curve, Subgroup analysis",
author = "Caixia Li and Gl{\"u}er, {Claus C.} and Richard Eastell and Dieter Felsenberg and Reid, {David M.} and Christian Roux and Ying Lu",
year = "2012",
month = "12",
day = "1",
doi = "10.1016/j.acra.2012.09.007",
language = "English",
volume = "19",
pages = "1529--1536",
journal = "Academic Radiology",
issn = "1076-6332",
publisher = "Elsevier USA",
number = "12",

}

TY - JOUR

T1 - Tree-structured Subgroup Analysis of Receiver Operating Characteristic Curves for Diagnostic Tests

AU - Li, Caixia

AU - Glüer, Claus C.

AU - Eastell, Richard

AU - Felsenberg, Dieter

AU - Reid, David M.

AU - Roux, Christian

AU - Lu, Ying

PY - 2012/12/1

Y1 - 2012/12/1

N2 - Rationale and Objectives: Multiple diagnostic tests are often available for a disease. Their diagnostic accuracy may depend on the characteristics of testing subjects. The investigators propose a new tree-structured data-mining method that identifies subgroups and their corresponding diagnostic tests to achieve the maximum area under the receiver-operating characteristic curve. Materials and Methods: The Osteoporosis and Ultrasound Study is a prospectively designed, population-based European multicenter observational study to evaluate state-of-the-art diagnostic methods for assessing osteoporosis. A total 2837 women underwent dual x-ray absorptiometry (DXA) and quantitative ultrasound (QUS). Prevalent vertebral fractures were determined by a centralized radiology laboratory on the basis of radiographs. The data-mining algorithm includes three steps: defining the criteria for node splitting and selection of the best diagnostic test on the basis of the area under the curve, using a random forest to estimate the probability of DXA being the preferred diagnostic method for each participant, and building a single regression tree to describe subgroups for which either DXA or QUS is the more accurate test or for which the two tests are equivalent. Results: For participants with weights ≤54.5 kg, QUS had a higher area under the curve in identifying prevalent vertebral fracture. For participants whose weights were >58.5 kg and whose heights were ≤167.5 cm, DXA was better, and for the remaining participants, DXA and QUS had comparable accuracy and could be used interchangeably. Conclusions: The proposed tree-structured subgroup analysis successfully defines subgroups and their best diagnostic tests. The method can be used to develop optimal diagnostic strategies in personalized medicine.

AB - Rationale and Objectives: Multiple diagnostic tests are often available for a disease. Their diagnostic accuracy may depend on the characteristics of testing subjects. The investigators propose a new tree-structured data-mining method that identifies subgroups and their corresponding diagnostic tests to achieve the maximum area under the receiver-operating characteristic curve. Materials and Methods: The Osteoporosis and Ultrasound Study is a prospectively designed, population-based European multicenter observational study to evaluate state-of-the-art diagnostic methods for assessing osteoporosis. A total 2837 women underwent dual x-ray absorptiometry (DXA) and quantitative ultrasound (QUS). Prevalent vertebral fractures were determined by a centralized radiology laboratory on the basis of radiographs. The data-mining algorithm includes three steps: defining the criteria for node splitting and selection of the best diagnostic test on the basis of the area under the curve, using a random forest to estimate the probability of DXA being the preferred diagnostic method for each participant, and building a single regression tree to describe subgroups for which either DXA or QUS is the more accurate test or for which the two tests are equivalent. Results: For participants with weights ≤54.5 kg, QUS had a higher area under the curve in identifying prevalent vertebral fracture. For participants whose weights were >58.5 kg and whose heights were ≤167.5 cm, DXA was better, and for the remaining participants, DXA and QUS had comparable accuracy and could be used interchangeably. Conclusions: The proposed tree-structured subgroup analysis successfully defines subgroups and their best diagnostic tests. The method can be used to develop optimal diagnostic strategies in personalized medicine.

KW - Classification and regression tree

KW - Personalized medicine

KW - Random forest

KW - Receiver-operating characteristic curve

KW - Subgroup analysis

UR - http://www.scopus.com/inward/record.url?scp=84868296686&partnerID=8YFLogxK

U2 - 10.1016/j.acra.2012.09.007

DO - 10.1016/j.acra.2012.09.007

M3 - Article

VL - 19

SP - 1529

EP - 1536

JO - Academic Radiology

JF - Academic Radiology

SN - 1076-6332

IS - 12

ER -