-------------------
Version: 1.2-2
Date: 2015-04-10
* Fixed bug when no eligible states for state entry and/or exit
distributions.
* Fixed bug for plotting state occupation probabilities when state
names are character not numeric.
-------------------
Version: 1.2-0
Date: 2015-02-26
* Fixed estimation when state dependent censoring is
present. Now provide two functions 'DS.ind' and 'DS.dep' for
implementing Datta-Satten estimator (for independent censoring and
state-dependent censoring, respectively). Also provided
illustration based on simulated data in updated vignette.
* Dropped the 'Cens' argument from 'DS.ind' and 'DS.dep'
(censoring state assumed to be '0' throughout)
* Fixed error in 'Pst' function so that time interval is now s <
time <= t (previously was s <= time <= t)
* The covariance matrix returned by 'Pst' is now for only a single
time point (time t)
* Modified 'Dist' function to only estimate entry / exit
distributions when appropriate. That is, for non-recurrent states
with a unique path (consisting only of that state) between the
states before and after the given state. Additionally, entry
distributions are not calculated for initial states and exit
distributions are not calculated for terminal states. This
required corresponding changes to a number of functions including
'msSurv', 'EntryExit', 'Dist.CIs', 'BS.var', and the 'summary' and
'plot' methods for msSurv objects.
* A warning is given if left-truncation is present but the LT=TRUE
argument to 'msSurv' is not specified
* Fixed the title for the normalized state exit plots
* Added 'Rgraphviz' to the 'Depends' field in the DESCRIPTION file
* Fixed sum of transitions out of state at a given time to exclude
censoring events. This caused issues in the variance calculations
of the AJ estimator when censored times were tied with event times.
-------------------
Version: 1.1-2
Date: 2012-09-18
* Now exporting accessor functions for 'msSurv'
* Changed license from GPL-3 to GPL-2
* Added citation to JSS publication for msSurv (JSS 50:14, 2012)
* Included modified JSS paper as package vignette
* Changed all occurrences of 'sum.dNs' to 'sum_dNs' to avoid
conflict with S3 generic 'sum'
* Added package 'KMsurv' to the 'Suggests' field in the
DESCRIPTION file
-------------------
Version: 1.1-1
Date: 2012-06-25
* Modified output of 'print', 'show', and 'EntryExit' slightly
-------------------
Version: 1.1-0
Date: 2012-04-18
* Changed name of function 'st.t' to 'SOPt'
* Changed name of internal function 'stocc' to 'AJ.estimator'
* Removed internal function 'Dist.BS.var'
* Modified all functions to make more extensive use of vectorized
calculations
* Output now returned invisibly for functions Pst, SOPt, and
EntryExit
* Added accessor functions for ALL slots in msSurv objects.
* Updated code throughout to use accessor functions rather than
make explicit reference to slot names.
* Updated functions throughout msSurv package so that the labels
accessed the state names instead of column / array index locations.
* Changed required names for data columns to 'id', 'start',
'stop', 'start.state', and 'end.state' ('start' is optional for
right-censored only data)
* Removed arguments 'start.states' and 'd.var' from function
msSurv.
* Added argument 'bs' to determine whether bootstrapping is
used to calculate the variance. Bootstrapping is currently needed
for the variance of the A-J estimators when there is dependent
censoring, for the variance of the state entry / exit functions,
and for the variance of the state occupation probabilities when
there is > 1 possible starting state.
* Starting proportions in each state (prior to the first observed
transition time) are now calculated empirically based on the data
* In class msSurv, changed names for the following slots: 'out' to
'cov.AJs', 'cov.p' to 'var.sop', 'all.ajs' to 'AJs', and
'all.I_dA' to 'I.dA'.
* Added the computation and appropriate applications and output
for non-normalized state entry and exit distributions (referred to
as subdistributions). These are now calculated and stored in the
'Fsub' and 'Gsub' slots, respectively, in msSurv objects. Options
for plotting subdistribution functions, and for calculating them
using the EntryExit function, have been added, along with
appropriately named accessor functions.
* Changed notation for normalized state entry and exit
distributions from 'Fs' and 'Gs' to 'Fnorm' and 'Gnorm'. Added
accessor functions Fnorm and Gnorm.
* Changed the normalization for the state exit distributions
('Gnorm') so that it matches that used for the state entry
distributions. That is, the denominator consists of all
individuals who will ever *enter*, not necessarily leave, a state.
* Added the 'norm=TRUE' argument to the Dist.CIs function to allow
users to specify whether they want normalized entry/exit
distributions or if they prefer the non-normalized
subdistributions.
* Added "->" to output for print and show methods to clearly
identify transitions between states.
* Added a check and corresponding error message when users request
'bs=TRUE' with fewer than 10 subjects
* Added a check to ensure that all start times are earlier than
the stop times
* Now requiring users to insert a " " between transitioning states
they want to plot in the 'plot' function to clearly distinguish
transitions, e.g. "1 12" instead of "112"
* Modified argument "plot.type" in plot method for msSurv objects
to include possibilities 'entry.sub', 'entry.norm', 'exit.sub', or
'exit.norm' for state entry and exit plots.
* Created a "State Entry and Exit Distribution" section of the
summary method and removed that information from the "State
Occupation Probability" section.
* Modified 'print' and 'show' methods to display sparse output
concerning the msSurv object, and 'summary' to give more detailed
information. Added arguments to the summary method to allow users to
customize displayed output, including options for displaying /
hiding the transition probabilities, state occupation
probabilities, state entry / exit distributions, and Datta-Satten
weighted estimates of the counting processes. Additionally, users
can specify at which timepoints they want summary information.
* Updated author contact information in the description file.