Who can help me with the modelling of the following:
1. simple situation:
A cohort of individuals receive an operation, after which they are subjected
to a steadily decreasing risk of post-operative death (risk-time
relationship could be modeled by a Weibull function, for example).
2. envisaged - more difficult - situation:
the same post-operative mortality applies, but individuals receive their
operations at variable time (because the time until they need the operation
varies).
3. a variant
the risk is not a mathematical function, but calculated from a set of
empirical survival data.
I am using ithink Analyst for my model.
Any help or hints towards literature will be highly appreciated.
--------------------------------
Dr. Franz von Roenne
34/35 Grand Parade
London N4 1AQ, UK
tel/fax *44 - 181 - 802 0428
roenne@netcomuk.co.uk
modelling survival - how?
-
- Member
- Posts: 21
- Joined: Fri Mar 29, 2002 3:39 am
modelling survival - how?
Its not clear from your message what your modeling problem is.
Empirical data from each of the situations you describe can be
analyzed using standard survival statistics methods to either
estimate the parameters of the survival distribution (or a
non-parametric survival distribution estimate) or to conduct
hypothesis tests (one or two sample). Then, using the estimated
survival distribution projected survival rates can be calculated. If
you have a simulation problem, what is it?
Bill Steinhurst
From: "William Steinhurst" <wsteinhu@psd.state.vt.us>
Empirical data from each of the situations you describe can be
analyzed using standard survival statistics methods to either
estimate the parameters of the survival distribution (or a
non-parametric survival distribution estimate) or to conduct
hypothesis tests (one or two sample). Then, using the estimated
survival distribution projected survival rates can be calculated. If
you have a simulation problem, what is it?
Bill Steinhurst
From: "William Steinhurst" <wsteinhu@psd.state.vt.us>
modelling survival - how?
Concerning modeling survival:
Here are a few ways to approximate the situation you describe (if I
understand it correctly).
1. You could have an aging chain where each level has two outflows:
One is the usual outflow into the next level in the chain. The other is
an outflow to a sink (i.e. out of the system-- death), formulated as a
decay i.e. Level*probability of dieing). Each level in the chain
would represent a group of people who have survived a given time since
they had the operation (determined by the usual time constants in the
aging chain). Each level would have its own probability of dieing.
2. You could use a coflow structure. The level would be the population
of patients. The characteristic would be the average time since each
member of the population had his or her operation. The outflow from the
population would be a function of this time-since-operation.
3. You could use a coflow in which the primary level again would be the
population of patients, but where the characteristic would be the
probability of death. This probability could decay at some steady rate.
4. You could use a cascaded coflow with option 3.
For more detail on these structures please send me an email or check out
the Molecules.
Regards,
Jim Hines
LeapTec and MIT
JimHines@interserv.com
Here are a few ways to approximate the situation you describe (if I
understand it correctly).
1. You could have an aging chain where each level has two outflows:
One is the usual outflow into the next level in the chain. The other is
an outflow to a sink (i.e. out of the system-- death), formulated as a
decay i.e. Level*probability of dieing). Each level in the chain
would represent a group of people who have survived a given time since
they had the operation (determined by the usual time constants in the
aging chain). Each level would have its own probability of dieing.
2. You could use a coflow structure. The level would be the population
of patients. The characteristic would be the average time since each
member of the population had his or her operation. The outflow from the
population would be a function of this time-since-operation.
3. You could use a coflow in which the primary level again would be the
population of patients, but where the characteristic would be the
probability of death. This probability could decay at some steady rate.
4. You could use a cascaded coflow with option 3.
For more detail on these structures please send me an email or check out
the Molecules.
Regards,
Jim Hines
LeapTec and MIT
JimHines@interserv.com