| Title: | Boosting Multistate Models |
|---|---|
| Description: | Contains infrastructure for using mboost::gamboost() in order to estimate multistate models. |
| Authors: | Holger Reulen |
| Maintainer: | Holger Reulen <[email protected]> |
| License: | GPL (>= 2) |
| Version: | 1.1.88 |
| Built: | 2025-03-10 04:03:45 UTC |
| Source: | https://github.com/cran/gamboostMSM |
Gradient boosting for Cox-type multi state models: minimization of negative partial log likelihood using component- and transition-wise base-learners.
This package provides function objects to fit Cox-type multi state models
using the functional gradient descent boosting algorithm as implemented in the
splendid package mboost. Therefore, function Family() for fitting
multi state models is given, including negative log partial likelihood of a
Cox-type multi state model as risk function and its negative first partial
derivative with respect to the linear predictor as working response function.
Holger Reulen
Andersen, P. K., Borgan, O., Gill, R. D., Keiding, N. (1993) Statistical Models Based on Counting Processes. Springer Series in Statistics, New York: Springer-Verlag.
Buehlmann, P. Hothorn, T. (2007) Boosting Algorithms: Regularization, Prediction and Model Fitting (with Discussion). Statistical Science, 22(4), p. 477–505.
Hothorn, T., Buehlmann, P., Kneib, T., Schmid, M., Hofner, B. (2012) mboost: Model-Based Boosting, R package version 2.2-0. http://CRAN.R-project.org/package=mboost.
Kneib, T., Hothorn, T., Tutz, G. (2009) Variable Selection and Model Choice in Geoadditive Regression Models. BIOMETRICS 65, p. 626–634.
Ridgeway, G. (1999) The state of boosting. Computing Science and Statistics 31, p. 172–181.
This function calculates the Breslow estimator for the cumulative baseline hazard rate, given fitted linear predictor values.
breslow(f, riskset, entry, exit, trans, event)breslow(f, riskset, entry, exit, trans, event)
f |
fitted linear predictor values |
riskset |
riskset list as generated by |
entry |
entry times. |
exit |
exit times. |
trans |
transition index. |
event |
observed event indicator. |
This function calculates the Breslow estimator for the cumulative baseline hazard rate, given fitted linear predictor values.
A list of length Q with each element including including elements
times |
a vector of observed event times, |
cbhr |
a vector of calculated cumulative hazard rate values. |
Holger Reulen
## Not run: breslow(f, riskset, entry, exit, trans, event)## Not run: breslow(f, riskset, entry, exit, trans, event)
Calculates risksets needed for using family multistate.
buildrisksets(entry, exit, trans, event, statusinfo)buildrisksets(entry, exit, trans, event, statusinfo)
entry |
a vector with entry times. |
exit |
a vector with exit times. |
trans |
a vector with transition types. |
event |
a vector with noncensoring event indicators. |
statusinfo |
a logical indicating if information on the calculation process should be printed. |
This function calculates riksets needed for family multistate.
A list of length 2 with elements Ci and Ri, each
vectors of length n.
Holger Reulen
Cross-validation for Boosting Multi-state Models.
cvriskMSM(m, d, id, formulaMSM, xlist, qlist, k, riskset)cvriskMSM(m, d, id, formulaMSM, xlist, qlist, k, riskset)
m |
... |
d |
... |
id |
... |
formulaMSM |
... |
xlist |
... |
qlist |
... |
k |
... |
riskset |
... |
...
...
Holger Reulen
This function calculates the degrees of freedom as part of the calculation of the Akaike Information Criterion (AIC).
degreesoffreedom(m, statusinfo)degreesoffreedom(m, statusinfo)
m |
a boosted multi state model. |
statusinfo |
a logical indicating if information on the calculation process should be printed. |
This function calculates the degress of freedom as part of the calculation of the Akaike Information Criterion.
A vector of length equal to the number of boosting iterations in the plugged in model object.
Holger Reulen
## Not run: degreesoffreedom(m, statusinfo)## Not run: degreesoffreedom(m, statusinfo)
...
helpfunctionmultistate1(x, ef)helpfunctionmultistate1(x, ef)
x |
... |
ef |
... |
...
Holger Reulen
## Not run: helpfunctionmultistate1(x, ef)## Not run: helpfunctionmultistate1(x, ef)
...
helpfunctionmultistate2(x, dummy)helpfunctionmultistate2(x, dummy)
x |
... |
dummy |
... |
...
Holger Reulen
## Not run: helpfunctionmultistate2(x, dummy)## Not run: helpfunctionmultistate2(x, dummy)
...
meancentering(d, x, q, x.name, q.name)meancentering(d, x, q, x.name, q.name)
d |
data set |
x |
covariate |
q |
transition type(s) |
x.name |
name of the covariate for pasting the new transition type specific covariate name |
q.name |
name of the transition type for pasting the new transition type specific covariate name |
...
...
Holger Reulen
This function implements a family for
fitting multistate models using mboost.
multistate(Ri, Ci)multistate(Ri, Ci)
Ri |
a list giving the individual (i.e., spell specific) risksets. |
Ci |
a list giving the indexes of risksets an individual spell is a part of (see page 213 in the book Generalized Additive Models by T.J. Hastie and R.J. Tibshirani for a description). |
This function implements a family for multistate models and
will be used inside the gamboost or glmboost function.
Functions to be used inside gamboost.
Holger Reulen
...
plloss(event, f, Ri)plloss(event, f, Ri)
event |
... |
f |
... |
Ri |
... |
...
...
Holger Reulen
Plot cross-validation for boosting multi-state models.
plotcvriskMSM(cvriskMSMobject, type)plotcvriskMSM(cvriskMSMobject, type)
cvriskMSMobject |
result from |
type |
should all stratified results be plotted ( |
...
...
Holger Reulen