Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: AIC and BIC to compare parametric and non-parametric survival models


From   Tom Trikalinos <ttrikalin@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: AIC and BIC to compare parametric and non-parametric survival models
Date   Tue, 19 May 2009 11:38:20 -0400

Ronan, thanks for your points.
We are much earlier -- still in the phase of parameterizing a natural
history discrete events simulation model and calibrating it.

t


On Tue, May 19, 2009 at 4:22 AM, Ronan Conroy <rconroy@rcsi.ie> wrote:
> On 18 Beal 2009, at 18:27, Tom Trikalinos wrote:
>
>>>>> To compare non-parametric and parametric survival
>>>>> analysis models, can I use the AIC and BIC?
>>>>> Specifically, I fit Cox PH models and exponential and
>>>>> weibull parametric regressions. It was pointed out to
>>>>> me that AIC & BIC-based comparisons may not be valid
>>>>> (because Cox uses partial likelihood).
>>>>>
>>>>> PS. I am performing survival analyses to inform a decision
>>>>> analysis. For this reason I strongly prefer to fit
>>>>> parametric models - will make life easier and restore the
>>>>> smile on me face.
>
>
> Comparing models for decision making involves assessing the validity and
> utility of the decisions they make. This is not a matter of model fit but of
> defining the characteristics of a desirable model.
>
> For example, if you are developing a model to assess acute chest pain, then
> the critical model errors are false negatives - errors that will result in
> someone with an evolving heart attack being sent home. On the other hand, in
> many screening situations the problem is false positives, who will place a
> burden on diagnostic services.
>
> Ideally, you need to assess your model in a fresh sample. While you can hold
> back some of your estimation sample for validation, this poses the problem
> that factors such as sampling bias, measurement bias etc are common to your
> estimation and validation samples.
>
> I would go for the performance indicators: sensitivity, specificity,
> positive and negative predictive value. Decision makers understand them.
>
> I have seen numerous statistically significant disease predictors that made
> no difference at all the clinical decision making. Indeed, useful predictors
> are far rarer than significant ones.
>
>
>
> Ronan Conroy
> =================================
>
> rconroy@rcsi.ie
> Royal College of Surgeons in Ireland
> Epidemiology Department,
> Beaux Lane House, Dublin 2, Ireland
> +353 (0)1 402 2431
> +353 (0)87 799 97 95
> +353 (0)1 402 2764 (Fax - remember them?)
> http://rcsi.academia.edu/RonanConroy
>
> P    Before printing, think about the environment
>
>
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index