MATLAB: How to stratify covariates in coxphfit function

coxphfitstratasurvival

Hi,

First I want to know how should I find out which covariates to stratify. I've been told to check the p-values in the stats after running coxphfit with no stratification. Those covariates which have p-value lower than 0.05 should be stratified. Is this correct?

Second, how do I actually stratify the covariates? The documentation says to offer name-value pair ('Strata', Gender) as input for the coxphfit, but what if my data is a matrix? Should I put 'Strata', X(:, 2) for example if I want to stratify the second covariate?

Best Answer

It is possible to specify a column for stratification, as in following example,

load('lightbulb.mat');
b = coxphfit(lightbulb(:,2),lightbulb(:,1), ...
'Strata',lightbulb(:,3))

Find more details about when stratify should be used in the documentation here. The model is based on the assumption that the baseline hazard function depends on time, t, but the predictor variables do not. This assumption is also called the proportional hazards assumption. When you have variables that do not satisfy this assumption, you can consider using two extensions of Cox proportional hazards model: the stratified Cox model and the Cox model with time-dependent variables.

If the variables that do not satisfy the PH assumption are categorizable, use the stratified Cox model. If the variables that do not satisfy the PH assumption are time-dependent variables, use the Cox model with time-dependent variables.

Related Solutions

MATLAB: How to use the output arguments of the COXPHFIT function in MATLAB 7.13 (R2011b) to evaluate the Cox proportional hazards model at new points

The example below illustrates the use of COXPHFIT.

Please note that the use of the STAIRS function when evaluating the model at a variety of responses in 'y'. This is because the estimated baseline cumulative hazard function (i.e., h(t) in the documentation) is not smooth. The function COXPHFIT is doing a type of maximum likelihood estimation. The likelihood is maximized if the cumulative hazard jumps at observed failure times and is flat otherwise.

% Number of observations
n = 30;
% Number of predictors
p = 5;
%Observed Data
X_observed = rand(n,p);
y_observed = rand(n,1);
%Get mean of X
Xbar = mean(X_observed);
%Get model
[b, ~, H] = coxphfit(X_observed,y_observed);
%Use model
%Make new observation
x_new = rand(1,p);
%Evaluate the model for a variety of y for this new, fixed observation
stairs( H(:,1), H(:,2) * exp((x_new - Xbar)*b));
title('Model');
%Plot estimated baseline survivor function for a variety of y
figure;
stairs( H(:,1), exp(-H(:,2)));
title('Survivor');

MATLAB: Are there functions in the Statistics Toolbox for survival analysis or multinomial linear models

There are several functions in the Statistics Toolbox for use with survival analysis. To get started, take a look at the demos "Analyzing survival or reliability data" for an overview, and "Fitting custom univariate distributions, part 2" for some more advanced information. You can find both demos under "Fitting Distributions to Data" in the Statistics Toolbox demos.

Functions useful with survival analysis include the following functions that allow data to be censored:

DFITTOOL - Distribution fitting tool

ECDF - Empirical cumulative distribution function, computes the Kaplan-Meier estimate of the cumulative distribution function

ECDFHIST - Histogram from ecdf output

EVFIT, EXPFIT, GAMFIT, LOGNFIT, MLE, MLECOV, NORMFIT, WBLFIT

- Distribution parameter estimation

KSDENSITY - Kernel-smoothing density estimate

PROBPLOT - Probability plot

There is a GLMFIT function for fitting generalized linear models, including ones with a binomial response variable.

As of MATLAB (7.3) R2006b, the function MNRFIT performs multinomial logistic regression.

There are various models used for multinomial response variables when the response categories are ordered. Some of these can be fit as a series of binomial fits. For example, one model calls for a series of binomial fits for ordered category j+1 or higher, given that you are in category j or higher. These types of models can be fit using the GLMFIT function.

Best Answer

Related Solutions

MATLAB: How to use the output arguments of the COXPHFIT function in MATLAB 7.13 (R2011b) to evaluate the Cox proportional hazards model at new points

MATLAB: Are there functions in the Statistics Toolbox for survival analysis or multinomial linear models

Related Question