A Bayesian neural network algorithm identifies patients in whom post-EVAR surveillance may be unnecessary

A. Karthikesalingam, O. Attallah, X. Ma, E. Choke, M.J. Brown, R.D. Sayers, R.J. Hinchliffe, P.J. Holt, M.M. Thompson

Research output: Contribution to journalMeeting abstract

Abstract

Lifelong surveillance is not cost-effective after endovascular aneurysm repair (EVAR), but is required to detect aortic complications which are fatal if untreated (type 1/3 endoleak, sac expansion, device migration). Aneurysm morphology determines the probability of aortic complications and therefore the need for surveillance, but existing analyses have proven incapable of identifying patients at sufficiently low risk to justify abandoning surveillance. This study aimed to improve the prediction of aortic complications, through the application of machine-learning techniques. Patients undergoing EVAR at 2 centres were studied from 2004–2010. Aneurysm morphology had previously been studied to derive the SGVI Score for predicting aortic complications. Bayesian Neural Networks were designed using the same data, to dichotomise patients into groups at low- or high-risk of aortic complications. Network training was performed only on patients treated at centre 1. External validation was performed by assessing network performance independently of network training, on patients treated at centre 2. Discrimination was assessed by Kaplan-Meier analysis to compare aortic complications in predicted low-risk versus predicted high-risk patients. 761 patients aged 75 +/− 7 years underwent EVAR in 2 centres. Mean follow-up was 36+/− 20 months. Neural networks were created incorporating neck angu- lation/length/diameter/volume; AAA diameter/area/volume/length/tortuosity; and common iliac tortuosity/diameter. A 19-feature network predicted aor- tic complications with excellent discrimination and external validation (5-year freedom from aortic complications in predicted low-risk vs predicted high-risk patients: 97.9% vs. 63%; p < 0.0001). A Bayesian Neural-Network algorithm can identify patients in whom it may be safe to abandon surveillance after EVAR. This proposal requires prospective study.
Original languageEnglish
Article numberA3
Pages (from-to)7
Number of pages1
JournalThe British Journal of Surgery
Volume101
Issue numberS2
DOIs
Publication statusPublished - 1 Mar 2014
Event48th Annual Scientific Meeting of the Vascular Society of Great Britain and Ireland - Manchester, United Kingdom
Duration: 27 Nov 201329 Nov 2013

Fingerprint

Aneurysm
Endoleak
Tics
Kaplan-Meier Estimate
Neck
Costs and Cost Analysis
Equipment and Supplies

Bibliographical note

Supplement: Abstracts of the 48th Annual Scientific Meeting of the Vascular Society of Great Britain and Ireland, 27–29 November 2013.

Cite this

Karthikesalingam, A., Attallah, O., Ma, X., Choke, E., Brown, M. J., Sayers, R. D., ... Thompson, M. M. (2014). A Bayesian neural network algorithm identifies patients in whom post-EVAR surveillance may be unnecessary. The British Journal of Surgery, 101(S2), 7. [A3]. https://doi.org/10.1002/bjs.9475
Karthikesalingam, A. ; Attallah, O. ; Ma, X. ; Choke, E. ; Brown, M.J. ; Sayers, R.D. ; Hinchliffe, R.J. ; Holt, P.J. ; Thompson, M.M. / A Bayesian neural network algorithm identifies patients in whom post-EVAR surveillance may be unnecessary. In: The British Journal of Surgery. 2014 ; Vol. 101, No. S2. pp. 7.
@article{fb77a137b3d34460acaf70177cf23f13,
title = "A Bayesian neural network algorithm identifies patients in whom post-EVAR surveillance may be unnecessary",
abstract = "Lifelong surveillance is not cost-effective after endovascular aneurysm repair (EVAR), but is required to detect aortic complications which are fatal if untreated (type 1/3 endoleak, sac expansion, device migration). Aneurysm morphology determines the probability of aortic complications and therefore the need for surveillance, but existing analyses have proven incapable of identifying patients at sufficiently low risk to justify abandoning surveillance. This study aimed to improve the prediction of aortic complications, through the application of machine-learning techniques. Patients undergoing EVAR at 2 centres were studied from 2004–2010. Aneurysm morphology had previously been studied to derive the SGVI Score for predicting aortic complications. Bayesian Neural Networks were designed using the same data, to dichotomise patients into groups at low- or high-risk of aortic complications. Network training was performed only on patients treated at centre 1. External validation was performed by assessing network performance independently of network training, on patients treated at centre 2. Discrimination was assessed by Kaplan-Meier analysis to compare aortic complications in predicted low-risk versus predicted high-risk patients. 761 patients aged 75 +/− 7 years underwent EVAR in 2 centres. Mean follow-up was 36+/− 20 months. Neural networks were created incorporating neck angu- lation/length/diameter/volume; AAA diameter/area/volume/length/tortuosity; and common iliac tortuosity/diameter. A 19-feature network predicted aor- tic complications with excellent discrimination and external validation (5-year freedom from aortic complications in predicted low-risk vs predicted high-risk patients: 97.9{\%} vs. 63{\%}; p < 0.0001). A Bayesian Neural-Network algorithm can identify patients in whom it may be safe to abandon surveillance after EVAR. This proposal requires prospective study.",
author = "A. Karthikesalingam and O. Attallah and X. Ma and E. Choke and M.J. Brown and R.D. Sayers and R.J. Hinchliffe and P.J. Holt and M.M. Thompson",
note = "Supplement: Abstracts of the 48th Annual Scientific Meeting of the Vascular Society of Great Britain and Ireland, 27–29 November 2013.",
year = "2014",
month = "3",
day = "1",
doi = "10.1002/bjs.9475",
language = "English",
volume = "101",
pages = "7",
journal = "The British Journal of Surgery",
issn = "0007-1323",
publisher = "John Wiley and Sons Ltd",
number = "S2",

}

Karthikesalingam, A, Attallah, O, Ma, X, Choke, E, Brown, MJ, Sayers, RD, Hinchliffe, RJ, Holt, PJ & Thompson, MM 2014, 'A Bayesian neural network algorithm identifies patients in whom post-EVAR surveillance may be unnecessary', The British Journal of Surgery, vol. 101, no. S2, A3, pp. 7. https://doi.org/10.1002/bjs.9475

A Bayesian neural network algorithm identifies patients in whom post-EVAR surveillance may be unnecessary. / Karthikesalingam, A.; Attallah, O.; Ma, X.; Choke, E.; Brown, M.J.; Sayers, R.D.; Hinchliffe, R.J.; Holt, P.J.; Thompson, M.M.

In: The British Journal of Surgery, Vol. 101, No. S2, A3, 01.03.2014, p. 7.

Research output: Contribution to journalMeeting abstract

TY - JOUR

T1 - A Bayesian neural network algorithm identifies patients in whom post-EVAR surveillance may be unnecessary

AU - Karthikesalingam, A.

AU - Attallah, O.

AU - Ma, X.

AU - Choke, E.

AU - Brown, M.J.

AU - Sayers, R.D.

AU - Hinchliffe, R.J.

AU - Holt, P.J.

AU - Thompson, M.M.

N1 - Supplement: Abstracts of the 48th Annual Scientific Meeting of the Vascular Society of Great Britain and Ireland, 27–29 November 2013.

PY - 2014/3/1

Y1 - 2014/3/1

N2 - Lifelong surveillance is not cost-effective after endovascular aneurysm repair (EVAR), but is required to detect aortic complications which are fatal if untreated (type 1/3 endoleak, sac expansion, device migration). Aneurysm morphology determines the probability of aortic complications and therefore the need for surveillance, but existing analyses have proven incapable of identifying patients at sufficiently low risk to justify abandoning surveillance. This study aimed to improve the prediction of aortic complications, through the application of machine-learning techniques. Patients undergoing EVAR at 2 centres were studied from 2004–2010. Aneurysm morphology had previously been studied to derive the SGVI Score for predicting aortic complications. Bayesian Neural Networks were designed using the same data, to dichotomise patients into groups at low- or high-risk of aortic complications. Network training was performed only on patients treated at centre 1. External validation was performed by assessing network performance independently of network training, on patients treated at centre 2. Discrimination was assessed by Kaplan-Meier analysis to compare aortic complications in predicted low-risk versus predicted high-risk patients. 761 patients aged 75 +/− 7 years underwent EVAR in 2 centres. Mean follow-up was 36+/− 20 months. Neural networks were created incorporating neck angu- lation/length/diameter/volume; AAA diameter/area/volume/length/tortuosity; and common iliac tortuosity/diameter. A 19-feature network predicted aor- tic complications with excellent discrimination and external validation (5-year freedom from aortic complications in predicted low-risk vs predicted high-risk patients: 97.9% vs. 63%; p < 0.0001). A Bayesian Neural-Network algorithm can identify patients in whom it may be safe to abandon surveillance after EVAR. This proposal requires prospective study.

AB - Lifelong surveillance is not cost-effective after endovascular aneurysm repair (EVAR), but is required to detect aortic complications which are fatal if untreated (type 1/3 endoleak, sac expansion, device migration). Aneurysm morphology determines the probability of aortic complications and therefore the need for surveillance, but existing analyses have proven incapable of identifying patients at sufficiently low risk to justify abandoning surveillance. This study aimed to improve the prediction of aortic complications, through the application of machine-learning techniques. Patients undergoing EVAR at 2 centres were studied from 2004–2010. Aneurysm morphology had previously been studied to derive the SGVI Score for predicting aortic complications. Bayesian Neural Networks were designed using the same data, to dichotomise patients into groups at low- or high-risk of aortic complications. Network training was performed only on patients treated at centre 1. External validation was performed by assessing network performance independently of network training, on patients treated at centre 2. Discrimination was assessed by Kaplan-Meier analysis to compare aortic complications in predicted low-risk versus predicted high-risk patients. 761 patients aged 75 +/− 7 years underwent EVAR in 2 centres. Mean follow-up was 36+/− 20 months. Neural networks were created incorporating neck angu- lation/length/diameter/volume; AAA diameter/area/volume/length/tortuosity; and common iliac tortuosity/diameter. A 19-feature network predicted aor- tic complications with excellent discrimination and external validation (5-year freedom from aortic complications in predicted low-risk vs predicted high-risk patients: 97.9% vs. 63%; p < 0.0001). A Bayesian Neural-Network algorithm can identify patients in whom it may be safe to abandon surveillance after EVAR. This proposal requires prospective study.

UR - http://onlinelibrary.wiley.com/doi/10.1002/bjs.9475/abstract

U2 - 10.1002/bjs.9475

DO - 10.1002/bjs.9475

M3 - Meeting abstract

VL - 101

SP - 7

JO - The British Journal of Surgery

JF - The British Journal of Surgery

SN - 0007-1323

IS - S2

M1 - A3

ER -