Neural network and logistic regression diagnostic prediction models for giant cell arteritis: Development and validation

Edsel B. Ing, Neil R. Miller, Angeline Nguyen, Wanhua Su, Lulu L.C.D. Bursztyn, Meredith Poole, Vinay Kansal, Andrew Toren, Dana Albreki, Jack G. Mouhanna, Alla Muladzanov, Mikaël Bernier, Mark Gans, Dongho Lee, Colten Wendel, Claire Sheldon, Marc Shields, Lorne Bellan, Matthew Lee-Wing, Yasaman MohadjerNavdeep Nijhawan, Felix Tyndel, Arun N.E. Sundaram, Martin W. Ten Hove, John Chen, Amadeo R. Rodriguez, Angela Hu, Nader Khalidi, Royce Ing, Samuel W.K. Wong, Nurhan Torun

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

Purpose: To develop and validate neural network (NN) vs logistic regression (LR) diagnostic prediction models in patients with suspected giant cell arteritis (GCA). Design: Multicenter retrospective chart review. Methods: An audit of consecutive patients undergoing temporal artery biopsy (TABx) for suspected GCA was conducted at 14 international medical centers. The outcome variable was biopsy-proven GCA. The predictor variables were age, gender, headache, clinical temporal artery abnormality, jaw claudication, vision loss, diplopia, erythrocyte sedimentation rate, C-reactive protein, and platelet level. The data were divided into three groups to train, validate, and test the models. The NN model with the lowest false-negative rate was chosen. Internal and external validations were performed. Results: Of 1,833 patients who underwent TABx, there was complete information on 1,201 patients, 300 (25%) of whom had a positive TABx. On multivariable LR age, platelets, jaw claudication, vision loss, log C-reactive protein, log erythrocyte sedimentation rate, headache, and clinical temporal artery abnormality were statistically significant predictors of a positive TABx (P≤0.05). The area under the receiver operating characteristic curve/Hosmer-Lemeshow P for LR was 0.867 (95% CI, 0.794, 0.917)/0.119 vs NN 0.860 (95% CI, 0.786, 0.911)/0.805, with no statistically significant difference of the area under the curves (P=0.316). The misclassification rate/false-negative rate of LR was 20.6%/47.5% vs 18.1%/30.5% for NN. Missing data analysis did not change the results. Conclusion: Statistical models can aid in the triage of patients with suspected GCA. Misclassification remains a concern, but cutoff values for 95% and 99% sensitivities are provided (https://goo.gl/THCnuU).

Original languageEnglish (US)
Pages (from-to)421-430
Number of pages10
JournalClinical Ophthalmology
Volume13
DOIs
StatePublished - Jan 1 2019

Fingerprint

Giant Cell Arteritis
Temporal Arteries
Logistic Models
Regression Analysis
Blood Sedimentation
C-Reactive Protein
Jaw Abnormalities
Headache
Blood Platelets
Biopsy
Neural Networks (Computer)
Diplopia
Triage
Statistical Models
Jaw
ROC Curve
Area Under Curve

Keywords

  • Giant cell arteritis
  • Logistic regression
  • Neural network
  • Ophthalmology
  • Prediction models
  • Rheumatology
  • Temporal artery biopsy

ASJC Scopus subject areas

  • Ophthalmology

Cite this

Ing, E. B., Miller, N. R., Nguyen, A., Su, W., Bursztyn, L. L. C. D., Poole, M., ... Torun, N. (2019). Neural network and logistic regression diagnostic prediction models for giant cell arteritis: Development and validation. Clinical Ophthalmology, 13, 421-430. https://doi.org/10.2147/OPTH.S193460

Neural network and logistic regression diagnostic prediction models for giant cell arteritis : Development and validation. / Ing, Edsel B.; Miller, Neil R.; Nguyen, Angeline; Su, Wanhua; Bursztyn, Lulu L.C.D.; Poole, Meredith; Kansal, Vinay; Toren, Andrew; Albreki, Dana; Mouhanna, Jack G.; Muladzanov, Alla; Bernier, Mikaël; Gans, Mark; Lee, Dongho; Wendel, Colten; Sheldon, Claire; Shields, Marc; Bellan, Lorne; Lee-Wing, Matthew; Mohadjer, Yasaman; Nijhawan, Navdeep; Tyndel, Felix; Sundaram, Arun N.E.; Ten Hove, Martin W.; Chen, John; Rodriguez, Amadeo R.; Hu, Angela; Khalidi, Nader; Ing, Royce; Wong, Samuel W.K.; Torun, Nurhan.

In: Clinical Ophthalmology, Vol. 13, 01.01.2019, p. 421-430.

Research output: Contribution to journalArticle

Ing, EB, Miller, NR, Nguyen, A, Su, W, Bursztyn, LLCD, Poole, M, Kansal, V, Toren, A, Albreki, D, Mouhanna, JG, Muladzanov, A, Bernier, M, Gans, M, Lee, D, Wendel, C, Sheldon, C, Shields, M, Bellan, L, Lee-Wing, M, Mohadjer, Y, Nijhawan, N, Tyndel, F, Sundaram, ANE, Ten Hove, MW, Chen, J, Rodriguez, AR, Hu, A, Khalidi, N, Ing, R, Wong, SWK & Torun, N 2019, 'Neural network and logistic regression diagnostic prediction models for giant cell arteritis: Development and validation', Clinical Ophthalmology, vol. 13, pp. 421-430. https://doi.org/10.2147/OPTH.S193460
Ing, Edsel B. ; Miller, Neil R. ; Nguyen, Angeline ; Su, Wanhua ; Bursztyn, Lulu L.C.D. ; Poole, Meredith ; Kansal, Vinay ; Toren, Andrew ; Albreki, Dana ; Mouhanna, Jack G. ; Muladzanov, Alla ; Bernier, Mikaël ; Gans, Mark ; Lee, Dongho ; Wendel, Colten ; Sheldon, Claire ; Shields, Marc ; Bellan, Lorne ; Lee-Wing, Matthew ; Mohadjer, Yasaman ; Nijhawan, Navdeep ; Tyndel, Felix ; Sundaram, Arun N.E. ; Ten Hove, Martin W. ; Chen, John ; Rodriguez, Amadeo R. ; Hu, Angela ; Khalidi, Nader ; Ing, Royce ; Wong, Samuel W.K. ; Torun, Nurhan. / Neural network and logistic regression diagnostic prediction models for giant cell arteritis : Development and validation. In: Clinical Ophthalmology. 2019 ; Vol. 13. pp. 421-430.
@article{a3749bea290949e3be58d766548deafc,
title = "Neural network and logistic regression diagnostic prediction models for giant cell arteritis: Development and validation",
abstract = "Purpose: To develop and validate neural network (NN) vs logistic regression (LR) diagnostic prediction models in patients with suspected giant cell arteritis (GCA). Design: Multicenter retrospective chart review. Methods: An audit of consecutive patients undergoing temporal artery biopsy (TABx) for suspected GCA was conducted at 14 international medical centers. The outcome variable was biopsy-proven GCA. The predictor variables were age, gender, headache, clinical temporal artery abnormality, jaw claudication, vision loss, diplopia, erythrocyte sedimentation rate, C-reactive protein, and platelet level. The data were divided into three groups to train, validate, and test the models. The NN model with the lowest false-negative rate was chosen. Internal and external validations were performed. Results: Of 1,833 patients who underwent TABx, there was complete information on 1,201 patients, 300 (25{\%}) of whom had a positive TABx. On multivariable LR age, platelets, jaw claudication, vision loss, log C-reactive protein, log erythrocyte sedimentation rate, headache, and clinical temporal artery abnormality were statistically significant predictors of a positive TABx (P≤0.05). The area under the receiver operating characteristic curve/Hosmer-Lemeshow P for LR was 0.867 (95{\%} CI, 0.794, 0.917)/0.119 vs NN 0.860 (95{\%} CI, 0.786, 0.911)/0.805, with no statistically significant difference of the area under the curves (P=0.316). The misclassification rate/false-negative rate of LR was 20.6{\%}/47.5{\%} vs 18.1{\%}/30.5{\%} for NN. Missing data analysis did not change the results. Conclusion: Statistical models can aid in the triage of patients with suspected GCA. Misclassification remains a concern, but cutoff values for 95{\%} and 99{\%} sensitivities are provided (https://goo.gl/THCnuU).",
keywords = "Giant cell arteritis, Logistic regression, Neural network, Ophthalmology, Prediction models, Rheumatology, Temporal artery biopsy",
author = "Ing, {Edsel B.} and Miller, {Neil R.} and Angeline Nguyen and Wanhua Su and Bursztyn, {Lulu L.C.D.} and Meredith Poole and Vinay Kansal and Andrew Toren and Dana Albreki and Mouhanna, {Jack G.} and Alla Muladzanov and Mika{\"e}l Bernier and Mark Gans and Dongho Lee and Colten Wendel and Claire Sheldon and Marc Shields and Lorne Bellan and Matthew Lee-Wing and Yasaman Mohadjer and Navdeep Nijhawan and Felix Tyndel and Sundaram, {Arun N.E.} and {Ten Hove}, {Martin W.} and John Chen and Rodriguez, {Amadeo R.} and Angela Hu and Nader Khalidi and Royce Ing and Wong, {Samuel W.K.} and Nurhan Torun",
year = "2019",
month = "1",
day = "1",
doi = "10.2147/OPTH.S193460",
language = "English (US)",
volume = "13",
pages = "421--430",
journal = "Clinical Ophthalmology",
issn = "1177-5467",
publisher = "Dove Medical Press Ltd.",

}

TY - JOUR

T1 - Neural network and logistic regression diagnostic prediction models for giant cell arteritis

T2 - Development and validation

AU - Ing, Edsel B.

AU - Miller, Neil R.

AU - Nguyen, Angeline

AU - Su, Wanhua

AU - Bursztyn, Lulu L.C.D.

AU - Poole, Meredith

AU - Kansal, Vinay

AU - Toren, Andrew

AU - Albreki, Dana

AU - Mouhanna, Jack G.

AU - Muladzanov, Alla

AU - Bernier, Mikaël

AU - Gans, Mark

AU - Lee, Dongho

AU - Wendel, Colten

AU - Sheldon, Claire

AU - Shields, Marc

AU - Bellan, Lorne

AU - Lee-Wing, Matthew

AU - Mohadjer, Yasaman

AU - Nijhawan, Navdeep

AU - Tyndel, Felix

AU - Sundaram, Arun N.E.

AU - Ten Hove, Martin W.

AU - Chen, John

AU - Rodriguez, Amadeo R.

AU - Hu, Angela

AU - Khalidi, Nader

AU - Ing, Royce

AU - Wong, Samuel W.K.

AU - Torun, Nurhan

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Purpose: To develop and validate neural network (NN) vs logistic regression (LR) diagnostic prediction models in patients with suspected giant cell arteritis (GCA). Design: Multicenter retrospective chart review. Methods: An audit of consecutive patients undergoing temporal artery biopsy (TABx) for suspected GCA was conducted at 14 international medical centers. The outcome variable was biopsy-proven GCA. The predictor variables were age, gender, headache, clinical temporal artery abnormality, jaw claudication, vision loss, diplopia, erythrocyte sedimentation rate, C-reactive protein, and platelet level. The data were divided into three groups to train, validate, and test the models. The NN model with the lowest false-negative rate was chosen. Internal and external validations were performed. Results: Of 1,833 patients who underwent TABx, there was complete information on 1,201 patients, 300 (25%) of whom had a positive TABx. On multivariable LR age, platelets, jaw claudication, vision loss, log C-reactive protein, log erythrocyte sedimentation rate, headache, and clinical temporal artery abnormality were statistically significant predictors of a positive TABx (P≤0.05). The area under the receiver operating characteristic curve/Hosmer-Lemeshow P for LR was 0.867 (95% CI, 0.794, 0.917)/0.119 vs NN 0.860 (95% CI, 0.786, 0.911)/0.805, with no statistically significant difference of the area under the curves (P=0.316). The misclassification rate/false-negative rate of LR was 20.6%/47.5% vs 18.1%/30.5% for NN. Missing data analysis did not change the results. Conclusion: Statistical models can aid in the triage of patients with suspected GCA. Misclassification remains a concern, but cutoff values for 95% and 99% sensitivities are provided (https://goo.gl/THCnuU).

AB - Purpose: To develop and validate neural network (NN) vs logistic regression (LR) diagnostic prediction models in patients with suspected giant cell arteritis (GCA). Design: Multicenter retrospective chart review. Methods: An audit of consecutive patients undergoing temporal artery biopsy (TABx) for suspected GCA was conducted at 14 international medical centers. The outcome variable was biopsy-proven GCA. The predictor variables were age, gender, headache, clinical temporal artery abnormality, jaw claudication, vision loss, diplopia, erythrocyte sedimentation rate, C-reactive protein, and platelet level. The data were divided into three groups to train, validate, and test the models. The NN model with the lowest false-negative rate was chosen. Internal and external validations were performed. Results: Of 1,833 patients who underwent TABx, there was complete information on 1,201 patients, 300 (25%) of whom had a positive TABx. On multivariable LR age, platelets, jaw claudication, vision loss, log C-reactive protein, log erythrocyte sedimentation rate, headache, and clinical temporal artery abnormality were statistically significant predictors of a positive TABx (P≤0.05). The area under the receiver operating characteristic curve/Hosmer-Lemeshow P for LR was 0.867 (95% CI, 0.794, 0.917)/0.119 vs NN 0.860 (95% CI, 0.786, 0.911)/0.805, with no statistically significant difference of the area under the curves (P=0.316). The misclassification rate/false-negative rate of LR was 20.6%/47.5% vs 18.1%/30.5% for NN. Missing data analysis did not change the results. Conclusion: Statistical models can aid in the triage of patients with suspected GCA. Misclassification remains a concern, but cutoff values for 95% and 99% sensitivities are provided (https://goo.gl/THCnuU).

KW - Giant cell arteritis

KW - Logistic regression

KW - Neural network

KW - Ophthalmology

KW - Prediction models

KW - Rheumatology

KW - Temporal artery biopsy

UR - http://www.scopus.com/inward/record.url?scp=85065293212&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85065293212&partnerID=8YFLogxK

U2 - 10.2147/OPTH.S193460

DO - 10.2147/OPTH.S193460

M3 - Article

AN - SCOPUS:85065293212

VL - 13

SP - 421

EP - 430

JO - Clinical Ophthalmology

JF - Clinical Ophthalmology

SN - 1177-5467

ER -