Assessing fracture risk using gradient boosting machine (GBM) models

Elizabeth J. Atkinson, Terry M Therneau, L. Joseph Melton, Jon J. Camp, Sara J. Achenbach, Shreyasee Amin, Sundeep Khosla

Research output: Contribution to journalArticle

25 Citations (Scopus)

Abstract

Advanced bone imaging with quantitative computed tomography (QCT) has had limited success in significantly improving fracture prediction beyond standard areal bone mineral density (aBMD) measurements. Thus, we examined whether a machine learning paradigm, gradient boosting machine (GBM) modeling, which can incorporate diverse measurements of bone density and geometry from central QCT imaging and of bone microstructure from high-resolution peripheral QCT imaging, can improve fracture prediction. We studied two cohorts of postmenopausal women: 105 with and 99 without distal forearm fractures (Distal Forearm Cohort) and 40 with at least one grade 2 or 3 vertebral deformity and 78 with no vertebral fracture (Vertebral Cohort). Within each cohort, individual bone density, structure, or strength variables had areas under receiver operating characteristic curves (AUCs) ranging from 0.50 to 0.84 (median 0.61) for discriminating women with and without fracture. Using all possible variables in the GBM model, the AUCs were close to 1.0. Fracture predictions in the Vertebral Cohort using the GBM models built with the Distal Forearm Cohort had AUCs of 0.82-0.95, whereas predictions in the Distal Forearm Cohort using models built with the Vertebral Cohort had AUCs of 0.80-0.83. Attempts at capturing a comparable parametric model using the top variables from the Distal Forearm Cohort resulted in resulted in an AUC of 0.81. Relatively high AUCs for differing fracture types suggest that an underlying fracture propensity is being captured by this modeling approach. More complex modeling, such as with GBM, creates stronger fracture predictions and may allow deeper insights into information provided by advanced bone imaging techniques.

Original languageEnglish (US)
Pages (from-to)1397-1404
Number of pages8
JournalJournal of Bone and Mineral Research
Volume27
Issue number6
DOIs
StatePublished - Jun 2012

Fingerprint

Area Under Curve
Forearm
Bone Density
Tomography
Bone and Bones
ROC Curve

Keywords

  • BONE DENSITY
  • BONE QUALITY
  • DISTAL FOREARM FRACTURE
  • GRADIENT BOOSTING
  • QCT
  • VERTEBRAL FRACTURE

ASJC Scopus subject areas

  • Orthopedics and Sports Medicine
  • Endocrinology, Diabetes and Metabolism

Cite this

Assessing fracture risk using gradient boosting machine (GBM) models. / Atkinson, Elizabeth J.; Therneau, Terry M; Melton, L. Joseph; Camp, Jon J.; Achenbach, Sara J.; Amin, Shreyasee; Khosla, Sundeep.

In: Journal of Bone and Mineral Research, Vol. 27, No. 6, 06.2012, p. 1397-1404.

Research output: Contribution to journalArticle

Atkinson, Elizabeth J. ; Therneau, Terry M ; Melton, L. Joseph ; Camp, Jon J. ; Achenbach, Sara J. ; Amin, Shreyasee ; Khosla, Sundeep. / Assessing fracture risk using gradient boosting machine (GBM) models. In: Journal of Bone and Mineral Research. 2012 ; Vol. 27, No. 6. pp. 1397-1404.
@article{462c5c599df648efb716ab9b633814e1,
title = "Assessing fracture risk using gradient boosting machine (GBM) models",
abstract = "Advanced bone imaging with quantitative computed tomography (QCT) has had limited success in significantly improving fracture prediction beyond standard areal bone mineral density (aBMD) measurements. Thus, we examined whether a machine learning paradigm, gradient boosting machine (GBM) modeling, which can incorporate diverse measurements of bone density and geometry from central QCT imaging and of bone microstructure from high-resolution peripheral QCT imaging, can improve fracture prediction. We studied two cohorts of postmenopausal women: 105 with and 99 without distal forearm fractures (Distal Forearm Cohort) and 40 with at least one grade 2 or 3 vertebral deformity and 78 with no vertebral fracture (Vertebral Cohort). Within each cohort, individual bone density, structure, or strength variables had areas under receiver operating characteristic curves (AUCs) ranging from 0.50 to 0.84 (median 0.61) for discriminating women with and without fracture. Using all possible variables in the GBM model, the AUCs were close to 1.0. Fracture predictions in the Vertebral Cohort using the GBM models built with the Distal Forearm Cohort had AUCs of 0.82-0.95, whereas predictions in the Distal Forearm Cohort using models built with the Vertebral Cohort had AUCs of 0.80-0.83. Attempts at capturing a comparable parametric model using the top variables from the Distal Forearm Cohort resulted in resulted in an AUC of 0.81. Relatively high AUCs for differing fracture types suggest that an underlying fracture propensity is being captured by this modeling approach. More complex modeling, such as with GBM, creates stronger fracture predictions and may allow deeper insights into information provided by advanced bone imaging techniques.",
keywords = "BONE DENSITY, BONE QUALITY, DISTAL FOREARM FRACTURE, GRADIENT BOOSTING, QCT, VERTEBRAL FRACTURE",
author = "Atkinson, {Elizabeth J.} and Therneau, {Terry M} and Melton, {L. Joseph} and Camp, {Jon J.} and Achenbach, {Sara J.} and Shreyasee Amin and Sundeep Khosla",
year = "2012",
month = "6",
doi = "10.1002/jbmr.1577",
language = "English (US)",
volume = "27",
pages = "1397--1404",
journal = "Journal of Bone and Mineral Research",
issn = "0884-0431",
publisher = "Wiley-Blackwell",
number = "6",

}

TY - JOUR

T1 - Assessing fracture risk using gradient boosting machine (GBM) models

AU - Atkinson, Elizabeth J.

AU - Therneau, Terry M

AU - Melton, L. Joseph

AU - Camp, Jon J.

AU - Achenbach, Sara J.

AU - Amin, Shreyasee

AU - Khosla, Sundeep

PY - 2012/6

Y1 - 2012/6

N2 - Advanced bone imaging with quantitative computed tomography (QCT) has had limited success in significantly improving fracture prediction beyond standard areal bone mineral density (aBMD) measurements. Thus, we examined whether a machine learning paradigm, gradient boosting machine (GBM) modeling, which can incorporate diverse measurements of bone density and geometry from central QCT imaging and of bone microstructure from high-resolution peripheral QCT imaging, can improve fracture prediction. We studied two cohorts of postmenopausal women: 105 with and 99 without distal forearm fractures (Distal Forearm Cohort) and 40 with at least one grade 2 or 3 vertebral deformity and 78 with no vertebral fracture (Vertebral Cohort). Within each cohort, individual bone density, structure, or strength variables had areas under receiver operating characteristic curves (AUCs) ranging from 0.50 to 0.84 (median 0.61) for discriminating women with and without fracture. Using all possible variables in the GBM model, the AUCs were close to 1.0. Fracture predictions in the Vertebral Cohort using the GBM models built with the Distal Forearm Cohort had AUCs of 0.82-0.95, whereas predictions in the Distal Forearm Cohort using models built with the Vertebral Cohort had AUCs of 0.80-0.83. Attempts at capturing a comparable parametric model using the top variables from the Distal Forearm Cohort resulted in resulted in an AUC of 0.81. Relatively high AUCs for differing fracture types suggest that an underlying fracture propensity is being captured by this modeling approach. More complex modeling, such as with GBM, creates stronger fracture predictions and may allow deeper insights into information provided by advanced bone imaging techniques.

AB - Advanced bone imaging with quantitative computed tomography (QCT) has had limited success in significantly improving fracture prediction beyond standard areal bone mineral density (aBMD) measurements. Thus, we examined whether a machine learning paradigm, gradient boosting machine (GBM) modeling, which can incorporate diverse measurements of bone density and geometry from central QCT imaging and of bone microstructure from high-resolution peripheral QCT imaging, can improve fracture prediction. We studied two cohorts of postmenopausal women: 105 with and 99 without distal forearm fractures (Distal Forearm Cohort) and 40 with at least one grade 2 or 3 vertebral deformity and 78 with no vertebral fracture (Vertebral Cohort). Within each cohort, individual bone density, structure, or strength variables had areas under receiver operating characteristic curves (AUCs) ranging from 0.50 to 0.84 (median 0.61) for discriminating women with and without fracture. Using all possible variables in the GBM model, the AUCs were close to 1.0. Fracture predictions in the Vertebral Cohort using the GBM models built with the Distal Forearm Cohort had AUCs of 0.82-0.95, whereas predictions in the Distal Forearm Cohort using models built with the Vertebral Cohort had AUCs of 0.80-0.83. Attempts at capturing a comparable parametric model using the top variables from the Distal Forearm Cohort resulted in resulted in an AUC of 0.81. Relatively high AUCs for differing fracture types suggest that an underlying fracture propensity is being captured by this modeling approach. More complex modeling, such as with GBM, creates stronger fracture predictions and may allow deeper insights into information provided by advanced bone imaging techniques.

KW - BONE DENSITY

KW - BONE QUALITY

KW - DISTAL FOREARM FRACTURE

KW - GRADIENT BOOSTING

KW - QCT

KW - VERTEBRAL FRACTURE

UR - http://www.scopus.com/inward/record.url?scp=84859440161&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84859440161&partnerID=8YFLogxK

U2 - 10.1002/jbmr.1577

DO - 10.1002/jbmr.1577

M3 - Article

C2 - 22367889

AN - SCOPUS:84859440161

VL - 27

SP - 1397

EP - 1404

JO - Journal of Bone and Mineral Research

JF - Journal of Bone and Mineral Research

SN - 0884-0431

IS - 6

ER -