Improving Cox survival analysis with a neural‐Bayesian approach
- 24 August 2004
- journal article
- research article
- Published by Wiley in Statistics in Medicine
- Vol. 23 (19) , 2989-3012
Abstract
In this article we show that traditional Cox survival analysis can be improved upon when supplemented with sensible priors and analysed within a neural Bayesian framework. We demonstrate that the Bayesian method gives more reliable predictions, in particular for relatively small data sets.The obtained posterior (the probability distribution of network parameters given the data) which in itself is intractable, can be made accessible by several approximations. We review approximations by Hybrid Markov Chain Monte Carlo sampling, a variational method and the Laplace approximation. We argue that although each Bayesian approach circumvents the shortcomings of the original Cox analysis, and therefore yields better predictive results, in practice the use of variational methods or Laplace is preferable. Since Cox survival analysis is infamous for its poor results with (too) many inputs, we use the Bayesian posterior to estimatep‐values on the inputs and to formulate an algorithm for backward elimination. We show that after removal of irrelevant inputs Bayesian methods still achieve significantly better results than classical Cox. Copyright © 2004 John Wiley & Sons, Ltd.Keywords
This publication has 20 references indexed in Scilit:
- Heuristics of instability and stabilization in model selectionThe Annals of Statistics, 1996
- Reversible jump Markov chain Monte Carlo computation and Bayesian model determinationBiometrika, 1995
- The restricted cubic spline as baseline hazard in the proportional hazards model with step function time‐dependent covariablesStatistics in Medicine, 1995
- Trees and splines in survival analysisStatistical Methods in Medical Research, 1995
- Penalized likelihood in Cox regressionStatistics in Medicine, 1994
- Survival analysis and neural netsStatistics in Medicine, 1994
- Assessing time‐by‐covariate interactions in proportional hazards regression models using cubic spline functionsStatistics in Medicine, 1994
- A technique for using neural network analysis to perform survival analysis of censored dataCancer Letters, 1994
- The Equivalence of Backward Elimination and Multiple ComparisonsJournal of the American Statistical Association, 1988
- Testing Precise HypothesesStatistical Science, 1987