A scalable discrete-time survival model for neural networks

Michael F. Gensheimer; Balasubramanian Narasimhan

doi:10.7717/peerj.6257

journal article Open Access Jan 25, 2019

A scalable discrete-time survival model for neural networks

Michael F. Gensheimer

Balasubramanian Narasimhan

PeerJ Vol. 7 pp. e6257 · PeerJ

View at Publisher Save 10.7717/peerj.6257

Abstract

There is currently great interest in applying neural networks to prediction tasks in medicine. It is important for predictive models to be able to use survival data, where each patient has a known follow-up time and event/censoring indicator. This avoids information loss when training the model and enables generation of predicted survival curves. In this paper, we describe a discrete-time survival model that is designed to be used with neural networks, which we refer to as Nnet-survival. The model is trained with the maximum likelihood method using mini-batch stochastic gradient descent (SGD). The use of SGD enables rapid convergence and application to large datasets that do not fit in memory. The model is flexible, so that the baseline hazard rate and the effect of the input data on hazard probability can vary with follow-up time. It has been implemented in the Keras deep learning framework, and source code for the model and several examples is available online. We demonstrate the performance of the model on both simulated and real data and compare it to existing models Cox-nnet and Deepsurv.

Topics

No keywords indexed for this article. Browse by subject →

References

25

[1]

Avati "Improving palliative care with deep learning" (2017) 10.1109/bibm.2017.8217669

[2]

Bottou "Stochastic gradient learning in neural networks" Proceedings of Neuro-Nimes (1991)

[3]

Breslow "Covariance analysis of censored survival data" Biometrics (1974) 10.2307/2529620

[4]

Breslow "A large sample study of the life table and product limit estimates under random censorship" The Annals of Statistics (1974) 10.1214/aos/1176342705

[5]

Brown "On the use of artificial neural networks for the analysis of survival data" IEEE Transactions on Neural Networks (1997) 10.1109/72.623209

[6]

Ching "Cox-nnet: an artificial neural network method for prognosis prediction of high-throughput omics data" PLOS Computational Biology (2018) 10.1371/journal.pcbi.1006076

[7]

Cooney "Value and limitations of existing scores for the assessment of cardiovascular risk: a review for clinicians" Journal of the American College of Cardiology (2009) 10.1016/j.jacc.2009.07.020

[8]

Regression Models and Life-Tables

D. R. Cox

Journal of the Royal Statistical Society Series B:... 1972 10.1111/j.2517-6161.1972.tb00899.x

[9]

Cox "Analysis of survival data" (1984)

[10]

A neural network model for survival data

David Faraggi, Richard Simon

Statistics in Medicine 1995 10.1002/sim.4780140108

[11]

Graf "Assessment and comparison of prognostic classification schemes for survival data" Stat Med (1999) 10.1002/(sici)1097-0258(19990915/30)18:17/18<2529::aid-sim274>3.0.co;2-5

[12]

Harrell Jr (2015) 10.1007/978-3-319-19425-7

[13]

Jackson "flexsurv: a platform for parametric survival modelling in R" Journal of Statistical Software (2016) 10.18637/jss.v070.i08

[14]

DeepSurv: personalized treatment recommender system using a Cox proportional hazards deep neural network

Jared L. Katzman, Uri Shaham, Alexander Cloninger et al.

BMC Medical Research Methodology 2018 10.1186/s12874-018-0482-1

[15]

Knaus "The SUPPORT prognostic model. Objective estimates of survival for seriously ill hospitalized adults" Annals of Internal Medicine (1995) 10.7326/0003-4819-122-3-199502010-00007

[16]

Kwong "A clinical score for predicting atrial fibrillation in patients with cryptogenic stroke or transient ischemic attack" Cardiology (2017) 10.1159/000476030

[17]

Gradient-based learning applied to document recognition

Y. Lecun, L. Bottou, Y. Bengio et al.

Proceedings of the IEEE 1998 10.1109/5.726791

[18]

Martinsson "WTTE-RNN: weibull time to event recurrent neural network" Master’s thesis (2016)

[19]

Deep Patient: An Unsupervised Representation to Predict the Future of Patients from the Electronic Health Records

Riccardo Miotto, Li Li, Brian A. Kidd et al.

Scientific Reports 2016 10.1038/srep26094

[20]

R Core Team "R: a language and environment for statistical computing" (2017)

[21]

Scalable and accurate deep learning with electronic health records

Alvin Rajkomar, Eyal Oren, Kang Chen et al.

npj Digital Medicine 2018 10.1038/s41746-018-0029-1

[22]

Rodriguez Lecture notes for WWS 509: generalized linear statistical models (2016) 10.1007/978-3-319-27637-3

[23]

Royston "External validation of a Cox prognostic model: principles and methods" BMC Medical Research Methodology (2013) 10.1186/1471-2288-13-33

[24]

Flexible parametric proportional‐hazards and proportional‐odds models for censored survival data, with application to prognostic modelling and estimation of treatment effects

Patrick Royston, Mahesh K. B. Parmar

Statistics in Medicine 2002 10.1002/sim.1203

[25]

It’s About Time: Using Discrete-Time Survival Analysis to Study Duration and the Timing of Events

Judith D. Singer, John B. Willett

Journal of Educational Statistics 1993 10.3102/10769986018002155

Metrics

182

Citations

25

References

Details

Published: Jan 25, 2019
Vol/Issue: 7
Pages: e6257
License: View

Authors

M

Michael F. Gensheimer

Department of Radiation Oncology, Stanford University, Stanford, CA, United States of America

B

Balasubramanian Narasimhan

Department of Statistics, Stanford University, Stanford, CA, United States of America

Cite This Article

Michael F. Gensheimer, Balasubramanian Narasimhan (2019). A scalable discrete-time survival model for neural networks. PeerJ, 7, e6257. https://doi.org/10.7717/peerj.6257

A scalable discrete-time survival model for neural networks

You May Also Like