Feed forward neural networks for the analysis of censored survival data

A partial logistic regression approach

Elia Biganzoli, Patrizia Boracchi, Luigi Mariani, Ettore Marubini

Research output: Contribution to journalArticle

140 Citations (Scopus)

Abstract

Flexible modelling in survival analysis can be useful both for exploratory and predictive purposes. Feed forward neural networks were recently considered for flexible non-linear modelling of censored survival data through the generalization of both discrete and continuous time models. We show that by treating the time interval as an input variable in a standard feed forward network with logistic activation and entropy error function, it is possible to estimate smoothed discrete hazards as conditional probabilities of failure. We considered an easily implementable approach with a fast selection criteria of the best configurations. Examples on data sets from two clinical trials are provided. The proposed artificial neural network (ANN) approach can be applied for the estimation of the functional relationships between covariates and time in survival data to improve model predictivity in the presence of complex prognostic relationships.

Original languageEnglish
Pages (from-to)1169-1186
Number of pages18
JournalStatistics in Medicine
Volume17
Issue number10
DOIs
Publication statusPublished - May 30 1998

Fingerprint

Censored Survival Data
Feedforward Neural Networks
Survival Analysis
Logistic Regression
Logistic Models
Partial
Feedforward Networks
Entropy Function
Nonlinear Modeling
Functional Relationship
Continuous-time Model
Discrete-time Model
Survival Data
Error function
Conditional probability
Hazard
Clinical Trials
Logistics
Artificial Neural Network
Covariates

ASJC Scopus subject areas

  • Epidemiology

Cite this

Feed forward neural networks for the analysis of censored survival data : A partial logistic regression approach. / Biganzoli, Elia; Boracchi, Patrizia; Mariani, Luigi; Marubini, Ettore.

In: Statistics in Medicine, Vol. 17, No. 10, 30.05.1998, p. 1169-1186.

Research output: Contribution to journalArticle

@article{b2c76632d8064efdae24f17c01c90e26,
title = "Feed forward neural networks for the analysis of censored survival data: A partial logistic regression approach",
abstract = "Flexible modelling in survival analysis can be useful both for exploratory and predictive purposes. Feed forward neural networks were recently considered for flexible non-linear modelling of censored survival data through the generalization of both discrete and continuous time models. We show that by treating the time interval as an input variable in a standard feed forward network with logistic activation and entropy error function, it is possible to estimate smoothed discrete hazards as conditional probabilities of failure. We considered an easily implementable approach with a fast selection criteria of the best configurations. Examples on data sets from two clinical trials are provided. The proposed artificial neural network (ANN) approach can be applied for the estimation of the functional relationships between covariates and time in survival data to improve model predictivity in the presence of complex prognostic relationships.",
author = "Elia Biganzoli and Patrizia Boracchi and Luigi Mariani and Ettore Marubini",
year = "1998",
month = "5",
day = "30",
doi = "10.1002/(SICI)1097-0258(19980530)17:10<1169::AID-SIM796>3.0.CO;2-D",
language = "English",
volume = "17",
pages = "1169--1186",
journal = "Statistics in Medicine",
issn = "0277-6715",
publisher = "John Wiley and Sons Ltd",
number = "10",

}

TY - JOUR

T1 - Feed forward neural networks for the analysis of censored survival data

T2 - A partial logistic regression approach

AU - Biganzoli, Elia

AU - Boracchi, Patrizia

AU - Mariani, Luigi

AU - Marubini, Ettore

PY - 1998/5/30

Y1 - 1998/5/30

N2 - Flexible modelling in survival analysis can be useful both for exploratory and predictive purposes. Feed forward neural networks were recently considered for flexible non-linear modelling of censored survival data through the generalization of both discrete and continuous time models. We show that by treating the time interval as an input variable in a standard feed forward network with logistic activation and entropy error function, it is possible to estimate smoothed discrete hazards as conditional probabilities of failure. We considered an easily implementable approach with a fast selection criteria of the best configurations. Examples on data sets from two clinical trials are provided. The proposed artificial neural network (ANN) approach can be applied for the estimation of the functional relationships between covariates and time in survival data to improve model predictivity in the presence of complex prognostic relationships.

AB - Flexible modelling in survival analysis can be useful both for exploratory and predictive purposes. Feed forward neural networks were recently considered for flexible non-linear modelling of censored survival data through the generalization of both discrete and continuous time models. We show that by treating the time interval as an input variable in a standard feed forward network with logistic activation and entropy error function, it is possible to estimate smoothed discrete hazards as conditional probabilities of failure. We considered an easily implementable approach with a fast selection criteria of the best configurations. Examples on data sets from two clinical trials are provided. The proposed artificial neural network (ANN) approach can be applied for the estimation of the functional relationships between covariates and time in survival data to improve model predictivity in the presence of complex prognostic relationships.

UR - http://www.scopus.com/inward/record.url?scp=0031921607&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0031921607&partnerID=8YFLogxK

U2 - 10.1002/(SICI)1097-0258(19980530)17:10<1169::AID-SIM796>3.0.CO;2-D

DO - 10.1002/(SICI)1097-0258(19980530)17:10<1169::AID-SIM796>3.0.CO;2-D

M3 - Article

VL - 17

SP - 1169

EP - 1186

JO - Statistics in Medicine

JF - Statistics in Medicine

SN - 0277-6715

IS - 10

ER -