Assessing fracture risk using gradient boosting machine (GBM) models

Oxford University Press (OUP) - Tập 27 Số 6 - Trang 1397-1404 - 2012
Elizabeth J. Atkinson1, Terry M. Therneau1, L. Joseph Melton2,3, Jon J. Camp4,5, Sara J. Achenbach1, Shreyasee Amin2,6, Sundeep Khosla3
1Division of Biomedical Statistics and Informatics, College of Medicine, Mayo Clinic, Rochester, MN, USA
2Division of Epidemiology, Department of Health Sciences Research, College of Medicine, Mayo Clinic, Rochester, MN, USA
3Divisions of Endocrinology, Metabolism, and Nutrition, College of Medicine, Mayo Clinic, Rochester, MN, USA
4Biomedical Imaging Resource
5Biomedical Imaging Resource; College of Medicine, Mayo Clinic, Rochester, MN, USA
6Division of Rheumatology, Department of Internal Medicine, College of Medicine, Mayo Clinic, Rochester, MN, USA

Tóm tắt

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. © 2012 American Society for Bone and Mineral Research.

Từ khóa


Tài liệu tham khảo

Genant, 2006, Advanced imaging assessment of bone quality, Ann N Y Acad Sci., 1068, 410, 10.1196/annals.1346.038

van Lenthe, 2006, CT-based visualization and quantification of bone microstructure in vivo, IBMS BoneKEy., 5, 410, 10.1138/20080348

Boutroy, 2005, In vivo assessment of trabecular bone microarchitecture by high-resolution peripheral quantitative computed tomography, J Clin Endocrinol Metab., 90, 6508, 10.1210/jc.2005-1258

Melton, 2007, Contribution of in vivo structural measurements and load/strength ratios to the determination of forearm fracture risk in postmenopausal women, J Bone Miner Res., 22, 1442, 10.1359/jbmr.070514

Melton, 2007, Structural determinants of vertebral fracture risk, J Bone Miner Res., 22, 1885, 10.1359/jbmr.070728

Sornay-Rendu, 2007, Alterations of cortical and trabecular architecture are associated with fractures in postmenopausal women, partially independent of decreased BMD measured by DXA: the OFELY Study, J Bone Miner Res., 22, 425, 10.1359/jbmr.061206

Boutroy, 2008, Finite element analysis based on in vivo HR-pQCT images of the distal radius is associated with wrist fracture in postmenopausal women, J Bone Miner Res., 23, 392, 10.1359/jbmr.071108

Vico, 2008, High-resolution pQCT analysis at the distal radius and tibia discriminates patients with recent wrist and femoral neck fractures, J Bone Miner Res., 23, 1741, 10.1359/jbmr.080704

Radspieler, 2008, In vivo assessment of 3-dimensional bone micro architecture with HR-pQCT in patients with and without fractures, J Bone Miner Res., 23, S310

Ladinsky, 2008, Trabecular structure quantified with the MRI-based virtual bone biopsy in postmenopausal women contributes to vertebral deformity burden independent of areal vertebral BMD, J Bone Miner Res., 23, 64, 10.1359/jbmr.070815

Sornay-Rendu, 2009, Severity of vertebral fractures is associated with alterations of cortical architecture in postmenopausal women, J Bone Miner Res., 24, 737, 10.1359/jbmr.081223

Hastie, 2009, The elements of statistical learning, data mining, inference, and prediction springer series in statistics, 2nd ed

Melton, 2010, Assessing forearm fracture risk in postmenopausal women, Osteoporos Int., 21, 1161, 10.1007/s00198-009-1047-2

Melton, 2010, Relation of vertebral fractures to bone density, structure and strength, J Bone Miner Res., 25, 1922, 10.1002/jbmr.150

Genant, 2003, Assessment of prevalent and incident vertebral fractures in osteoporosis research, Osteoporosis Int., 14, S43, 10.1007/s00198-002-1348-1

Seeman, 2006, Bone quality—the material and structural basis of bone strength and fragility, N Engl J Med., 354, 2250, 10.1056/NEJMra053077

ISCD

Kanis, 1994, The diagnosis of osteoporosis, J Bone Miner Res., 9, 1137, 10.1002/jbmr.5650090802

Silva, 1994, Direct and computed tomography thickness measurements of the human, lumbar vertebral shell and endplate, Bone., 15, 409, 10.1016/8756-3282(94)90817-6

Prevrhal, 2003, Accuracy of CT-based thickness measurement of thin structures: modeling of limited spatial resolution in all three dimensions, Med Phys., 30, 1, 10.1118/1.1521940

Kirmani, 2009, Bone structure at the distal radius during adolescent growth, J Bone Miner Res., 24, 1033, 10.1359/jbmr.081255

Laib, 1998, In vivo high resolution 3D-QCT of the human forearm, Technol Health Care., 6, 329, 10.3233/THC-1998-65-606

Parfitt, 1987, Bone histomorphometry: standardization of nomenclature, symbols, and units. Report of the ASBMR Histomorphometry Nomenclature Committee, J Bone Miner Res., 2, 595, 10.1002/jbmr.5650020617

Laib, 2002, New model-independent measures of trabecular bone structure applied to in vivo high-resolution MR images, Osteoporos Int., 13, 130, 10.1007/s001980200004

Laib, 1999, Calibration of trabecular bone structure measurements of in vivo three-dimensional peripheral quantitative computed tomography with 28-mm-resolution microcomputed tomography, Bone., 24, 35, 10.1016/S8756-3282(98)00159-8

MacNeil, 2007, Accuracy of high-resolution peripheral quantitative computed tomography for measurement of bone quality, Med Eng Phys., 29, 1096, 10.1016/j.medengphy.2006.11.002

Ridgeway, 2007, Generalized Boosted Regression Models

DeLong, 1988, Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach, Biometrics., 44, 837, 10.2307/2531595

Schapire, 1990, The strength of weak learnability, Machine Learn., 5, 197, 10.1007/BF00116037

Friedman, Greedy function approximation: a gradient boosting machines, IMS.

Friedman, 2002, Stochastic gradient boosting, Comput Stat Data Anal., 38, 367, 10.1016/S0167-9473(01)00065-2

Friedman, 2003, Multiple additive regression trees with application in epidemiology, Stat Med., 22, 1365, 10.1002/sim.1501

Elith, 2008, A working guide to boosted regression trees, J Anim Ecol., 77, 802, 10.1111/j.1365-2656.2008.01390.x

Hjort, 2008, Periglacial distribution modelling with a boosting method, Permafrost Periglacial Process., 20, 15, 10.1002/ppp.629

Jalabert, 2010, Estimating forest soil bulk density using boosted regression modelling, Soil Use Manage., 26, 516, 10.1111/j.1475-2743.2010.00305.x

Thông tin
Thông tin xuất bản

Oxford University Press (OUP)

Tập 27 Số 6

1397-1404

Thông tin tác giả