v2.6.3 (official release)
     - added the tests for carryover effects and the direct formulation effects with
       lm.mod() for 2x2x2 crossover study;
     - took care of switching pAUC to be FALSE when running 'statistical analysis only';
       since the demo dataset for 'statistical analysis only' does not include any
       pAUC parameter (demoBANOVA());
     - fixed BANOVA() again to simplify the codes;
     - started working with new release of R v3.0.3;
     - fixed BANOVAcsv() for dataset input if doing pAUC;
     - demo dataset do not have any pAUC example for statistical-only runs;
     - BANOVA() for pAUC stats. and 2x2x2 crossover with multiple-dose stats.;
     - fixed default values for 'pAUC_start' and 'pAUC_end' with '121' & '128',
       instead of '1' & '8' respectively; since the 'pAUC_start' & 'pAUC_end' 
       counts from time zero, not from one dosing interval, especially for multiple-dose;
       otherwise, it will cause error, such as 'Na/NaN/Inf -> y...'
     - fixed NCAplot() hopefully to speed-up plotting and enhance efficiency; 
     - changed parallel study in NCAoutput() as lme(); in RepMix() the parallel
       uses lm(); but the results are the same.

v2.6.2 (official release)
    - added lme_lm.mod() to simplify original RepMix();
    - added lm.mod() to simplify original BANOVA(); also fixed an error in original 
      BANOVA(); btw, fixed the output format of BANOVA();
    - fixed BANOVAanalyze() to run BANOVAoutput() once and RepMixanalyze() to run
      RepMIXOutput() once; no more run 'BANOVA() -> BANOVAoutput()' since BANOVAoutput()
      will call BANOVA(); and also no more 'RepMix() -> RepMixOutput()' since
      RepMIXoutput() will call RepMIX(); more efficient now; otherwise; BANOVA() & RepMix()
      will run/be called twice.
    - fixed some typo in the display of setting guide;
    - re-coded with NCAcsv(), BANOVAcsv() & NCA.BANOVAcsv(); more efficient hopefully;
    - excluded the parallel study from icd_check(); how can I forget this? and kept 
      the replicated crossover study; 
    - added 'input data' to description_version(); it only works for Windows OS;
    - added 'input dataset' and the name of OS platform to NCAoutput();
    - fixed icd_check() and abandoned using 'stop()' from it; icd.check()
      can be detected icd earlier than previous version since it will get
      work done once the dataset is loaded; only for nca data; not for anova
      dataset.
    - added 'run.demo','study.type' and 'dose.type' into setting items to
      reduce menu levels;
    - fixed 'dosing interval' in bear setting; forgot to change in the last
      release (v2.6.1); so this can cause error for multiple-dose;
    - disable dataset output for demo runs; all can be done from the top menu;
    - fixed CV_inter of BANOVA() when occurring negative variance components;
    - corrected typo in RepMix() to show CV(intra);
    - also fix NCAoutput() as adding ODA setting function;
    - integrate ODA into setting file; modify the top menu;
    - changed bear setup cheatsheet mode as a graphic display;
    - fixed setting display screen;
    - fixed output format for RepMIX();
    - re-assigned column names for 'statistical analysis' (BANOVA()) if the imported
      .csv was not originally generated by bear;
    - change RepMIX()'s lme() with lmeControl(opt='optim') since v2.6.2 since the
      default 'nlminb' fails to converge frequently;
    - fixed the inconsistency of 90% CI for replicate crossover in 'stat_sum_output' 
      and 'lme_output' with NCAoutput() and RepMIX();
    
v2.6.1 (official release)
    - added function to avoid the errors due to dummy input;
    - re-organize the information about how to setup bear using file.show();
      testing on Windows, linux and iMac OS X platforms; iMac will work with 
      R.app or RStudio,but not TK GUI;
    - added license (GPL-2|GPL-3) announcement for bear on all output or display 
      pages;
    - added the output function of individual data points (IDP output) with mean 
      & sd for each time point;
    - removed formfeed with NCAoutput(); it does not print correctly with a 
      pdf printer; don't know why.
    - fixed on-screen display content when it is pAUC;
    - fixed some typo;
    
v2.6.0 (official release)
    - decided not to implement data point interpolation for missing data value (C*)
      when doing partial AUC or truncated AUC analysis in bear; users may consider 
      to use an estimate/interpolation/extrapolation for missing data when 
      necessary beforehand; further ref. included:
      (1) http://forum.bebac.at/mix_entry.php?id=4655#p4681
      (2) http://forum.bebac.at/mix_entry.php?id=1933&page=0&category=0&order
                 =last_answer&descasc=DESC#p2323
    - added analysis for pAUC or truncated AUC (all noted as pAUC);
    - fixed AUMC to comply with linear-up/log-down trapezoidal rule as AUC.
    - added "study tracer" (footprint) following the selected items from menu;
    - create the export module for demo dataset;
    - enable demo dataset export through NCA and ANOVA for later testing purposes;
    - add 'plot.setup.rds'; plus original 'bear.setup.rds' of setup file for
      bear;
    - fixed export/import .csv and .Rdata file for 'statistical analysis' in 
      NCAoutput();
    - added bear.setup() for setup files generation; new features.
    - moved AUC calc. method to NCA.BANONAanalyze(); this has been changed later;
    - corrected some typos again;
    - fixed BANOVA() with no "abs()" for CV_intra values;
    - fixed descriptionTOST() using paste() for better display;
    
v2.5.8 (official release)
    - fixed BE criteria with percentage (/100); sorry about this error;
    - disable warning message with options(warn=-1) in go2menu();
    - add the function of NCA saved pivotal parameters back again with automatically
      saving as .RData (NCAoutput()); it was disabled before;
    - added more output info of study design for NCAoutput(); not just multiple-
      dose or single-dose; now also include if it is parallel or if it is replicated;
    - added LL/UL as data.frame() input for replicated study as the non-replicated;
    - fixed AIC crashed when running multiple-dose study; and
    - fixed plots of regression lines (ARS, AIC, etc.) for hiding the first
      empty (null) plot window.

v2.5.7 (official release)
    - official release on Sept. 27, 2013

v2.5.6.9  (pre-release)
-------------------------
    - added LL/UL as data.frame() input; not just LL;
    - fixed abnormal on-screen text display with Mac OS X using data.frame() input;
    - fixed BANOVAdata() & BANOVAcsv() with graphic menu; and also spelling check; and
    - removed abs() from CV_intra calculation (07/16/2013)

v2.5.6 (official release)
-------------------------
    - official release on June 27, 2013

v2.5.5.9 (pre-release)
-------------------------
    - added csv outputs for data point selections of linear regression to calculate
      lambda_z with ARS, AIC, TTT, TTTAIC & TTTARS now; it used to be only available
      for manual selection of data points; It's most easy function to implement so far
      which I cannot figure out how to do this long time ago.
    - fixed R^2 display on linear regression plots for terminal elim. rate constant (lambda_z)
      estimation; just a mathematical annotation (squared 2), it took me two whole
      nights to complete; Wow, it's not easy for this little thing...
    - fixed NCAselect() & NCAregplot() with plot(...,lab=c(15,15,40),...) for better
      looking.
    - fixed selection of all linear or lin-up/log-down calc. for AUC with manual selection
      of data points; others (AIC, ARS, TTT & etc. have been fixed last time); previously
      the selection does not work effectively, even the 'all linear' is selected.
    - corrected some typos; Is it cont. or cont'd? Answer: In formal writing use cont'd.
      ref. link: http://wiki.answers.com/Q/Is_it_cont_or_cont%27d (used in NCAOutput). 
      fine.
    - fixed 'description_pointselect()' to give more details about Windows, Linux/unix
      or ubuntu, and Mac OS X.
    - added lin-up/log-down trapezoidal AUC calc; the in-up/log-down trapezoidal method
      is default method now;
    - fix graphics in linux/unix (ubuntu), Mac OS X; use 'dev.new()' instead of 
      'windows(record=TRUE)'; original codes won't work any more for R v3.0 or above;
    - set options(digits=5) as default;
    
v2.5.5 (official release)
-------------------------
    - official release on May 07, 2013
    
v2.5.4.9 (pre-release)
-------------------------
    - OK, let's back to v2.5.4.9 as pre-release version; it makes more logical and
      convenient for users to upgrade from CRAN. The next official release will be
      v2.5.5, of course. Users only need to update their packages from CRAN in this case.
    - NCA plots are switched from plot() to ggplot() now; better graphics looking
      I think; add 'ggplot2' as Depends and 'plyr' as Suggests in DESCRIPTION.
    - use some helper functions from "Cookbook for R" authored by Winston Chang to
      use ggplot() (website);
    - remove console plotting now; all plots will be logged into .pdf of NCA plots;
      this will remove 5 R scripts from previous version;
    - add test or demo functions; split from original /R directory;
    - add page break for xxx_nca_outputs.txt;
    - fix bugs for the section containing the means (SD) for each formulation 
      (by time point) in NCA output file. (thanks to Elba Romero). This is only
      for the situations of running NCA demo or NCA.
    - improve data point selection methods with correct subject labeling;
    - add bear cover page for nca plot and ODA plot pdf files (using library(png));
    - fixed the part of data point manual selection logged as .pdf. With R3.0.0, it
      is still a problem; however, it is OK with R2.15.3 (miss it so much). Porbably
      in R3.0.0 the graphic windows have been changed as stated in its update doc. Now
      it should be fine. Quite different from ARS, AIC, TTT, etc.. my mistake.
    - remove the function to save 'TotalData' again after NCA as a .RData. It does not
      make any sense to do this.
    - remove DOA on-screen plots since all plots have been logged into a .pdf file.
      It's a redundancy procedure, just like on-screen NCA plots.
    - improve pivotal_output.csv; now it is much better than previously.
    - now manual data selection from previously saved .RData can also be plotted into
      pdf now, as well as on-screen plots.
    - fix 'Tmax_ss' for multiple dose BE; the original one should be subtracted with
      TlastD (the time of the final dose be given)
    - fix NCAselect again; try to improve graphic looking for multiple dose BE dataset,
      especially the setting of ylim(); done.
    - fixed the section containing the means (SD) for each formulation (by time point) 
      in NCA output file; it was the same for the ref. & test products in the previous
      version. Sorry about this.
    - hide on-screen linear regression plots, but still save all plots as one .pdf file.
    - set 'ylim=c(1e-3, 1e+5)' in all linear regression plots for lambda_z estimations; it
      looks better in this case.
    - all selection menu -> 'graphics=TRUE' now
    - 
      
v2.5.4
    - mostly based on BE/BA forums (bear feature request).
    - grouping each run with random batch# and system date marked; more outputs from this version. 
      and no more overwrite previous outputs now.
    - outlier detection analysis (ODA) is taken away from routine run; and also ODA outputs
      is separated;
    - max. data point selections for lambda_z estimation was increased to 6 (2-6);
    - single plot for point selection now (used to be 2x2 plots), old-man version?
    - fix inconsistent outputs for unbalanced dataset;
    - incomplete dataset will be blocked before being analyzed (hopefully);
    - use "file.choose()" for data file loading;
    - total lambda_z selections outputs as .csv format (sarawuto); not for demo functions;
    - pivotal and misc pk parameters output as .csv format;
    - change data() for demo functions (demodemu.r & demomenu1.r); to adapt future R (with R v3.0 
      alpha);
    - add "extdata" directory to store demo data files as .rda format;
    - add 'legend=FASLE' into lineplot.CI(...) (NCAplot.r) otherwise, lineplot.ci() package will 
      crash with error... [Error in rep_len(col, n.leg) : cannot replicate NULL to a non-zero 
      length] only with R v3.0.0 alpha;
    - add linear regression line plots for lambda_z estimation for manual data points selections (
      OK for R 2.15.3 and still got dev.copy() error when testing on R3.0.0beta (Helmut & sarawuto).
    - change save() and load() to saveRDS() and readRDS() for convenient scripting.
    - add mean (sd) plots on semilog scale (Helmut).
    - add alarm() and readline("...") for methods of estimation lambda_z to alert user what 
      they should know before next step.(as suggested by Detlew)
    - add a section containing the means (SD) for each formulation (by time point) in NCA output 
      file (as requested by Helmut).
    - modify some output files.
    - add plots of linear regression lines for lambda_z estimation for ARS, AIC, TTT,TTT-AIC &
      TTT-ARS.
    - change all load() and save() to readRDS() and saveRDS().
    
v2.5.3
    - generalized SAS-like outputs (such as ANOVA and Type III SS) with 2x2x2 crossover;
    - added variance model into lme() and fixed random effect in lme() for replicate study
      as suggested by D. Labes posted at bebac forum (browse this thread at bebac forum);
    - the maximum data points that can be chosen to estimation of fz is set to 4, at 
      least 2 data points are required. The "Select the exact 3 data points" has been changed 
      to "Select 2-4 data point"; and
    - other minor changes.
v2.5.2
    - changed anova model back to lm(log(PK)~ seq + subj:seq + prd + drug , data) as with 
      v2.4.0; if not, this will result in error in calculation of 90% CI, as well as point
      estimates (thanks to Ji?i Hofmann, Czech Republic).
v2.5.1
    - re-compiled again to change the requirement with R v2.10.0 to enable installation in MacOSX;
      however, bear was still compiled with R v2.11.0.
v2.5.0
    - added Welch t tests (also 90% CI) for a parallel BE study (thanks to Helmut Schutz);
    - added CVintra (intra-subject CV) calculation for a replicate BE study (lme_stat.txt)
      right after classical 90% CI;
    - hide ln(PK) list in the report of ANOVA_stat.txt/lm_stat.txt/lme_stat.txt;
    - completed the function of a parallel BE study with multiple-dosed (used to be single-dosed only);
      and
    - some other bugs fixed
v2.4.4
    - fixed the output file, Statistical_summaries.txt; i.e., Table 2: remove n1, n2 list from
      non-replicate or replicate crossover BE study; They only appear in Parallel BE study.
v2.4.3
    - changed the statistical model from linear mixed model(lme()) to linear model (lm())(similar to 
      SAS GLM) for a parallel BE study; basically, the results obtained from lme() or lm() are the same.
      However, to meet the regulatory requirements (both FDA and EU), lm() has been adopted since this
      version. Later, the Welch t-test will be added for a parallel BE study for unequal variances in
      the next release. See more discussion: Post#01 and Post#02.
    - recompiled bear under R v.2.11.0.
v2.4.2
    - fixed NCA outputs for the unbalanced parallel BE study
    - fixed TOST descriptions
    - recompiled with R v2.10.1
v2.4.1
    - add time/date stamp on the header or title page of text or pdf (plots) outputs (11.10(e) of 21 CRF 
      Part 11); more to be made for Part 11...
    - set .pdf output format as A4 paper size.
    - modify all .Rd files to fit R v2.9.2 requirements
    - change of the name "seq:subj" to "subj(seq)" in anova with the mode of "Y ~ seq + subj:seq +
      per + trt" in order to be conveniently cross validated with SAS (thanks to Elmaestro)
    - mostly minor changes with this release
v2.4.0
    - add data analysis of multiple-dosed (MD) ABE data
    - add Cook's distance for outlier detection with various criteria
    - fixed y-axis scaling problem
v2.3.1
    - fixed NCA plots of Time scale (with auto-scaled)
    - used the difference of lsmeans between the Ref. and the Test to calculate 90% CIs for 
      Cmax and AUCs
v2.3.0
    - add sample size estimation and lme analysis for the parallel BE study
    - add References into bear's output (such as ANOVA)
    - label outliers' subjects number for boxplot only if there is any (see crossover demo)
    - add input data summary of BA measurement (class level information, means, etc.)
    - add interpretations for some statistical tests (such as Hotelling T2 test)
    - add add Two One-Sided Tests (TOST) and Anderson-Hauck's tests (just for educational 
      purposes ONLY)
    - allow users to change BE acceptance range now (not fixed on 80%-125% any more!); different
      countries have different regulatory basis...
    - show MSResidual and MSSubject(seq) values when calculating inter- and intra-subject CV
    - change Hotelling T2 test layout

v 2.2.0
    - add point estimate along with 90% CI in output file called 'Statistical_summaries.txt 
      (suggested by D. Labes).
    - add sample size estimation for replicate BE study, and output re-arrangement.
    - add Hotelling T2 function and boxplot for outlier detection
    - add Quantiles for intrasubject and intersubject (with boxplot)
    - add replicated study for 2*2*3, 2*2*4, 2*2*5, and 2*2*6 (using lme to analyze replicated
      BE study)
    - add sample size estimation for replicated study (using 2*2*2 sample size estimation extended 
      to 2*2*n sample size of replicate crossover design)

v.2.1.0
    - we add 'analysis of outliers detection' since this release.  These include some normality 
      tests, and some diagnostic plots for this functions, such as QQ plots and intra- and 
      inter-subject residual plots.
    - fix intra-subject CV calculation based on ref. #6. (thanks to Helmut Schutz).

v1.5.0~2.0.3
   - now fz can be estimated from three methods: manual selections of the 3 exact data points,
     computer selection based on adjusted R sq. (ARS), and the Two-Times-Tmax (TTT) Method.
   - re-structure all codes of bear since v2.0.0.
   - add R sq. in NCA output; the original one is adjusted R sq. & changed T1/2 to T1/2(z) in NCA 
     output file
   - corrected some typo errors appearing in the output files or console display
   - rearrange the output files with better styles: such as 3-decimal digits for 'power', etc.
   - built-in the data file (.RData) required for demo purposes; user doesn't need to enter/import
     /load it again
   - manual selection of the 3 exact data points can be saved now.  user does not have to redo it 
     next time.
   - displayed the method used to estimate fz in NCA output
   - changed '0.693' to ln(2) when estimating T1/2(z) (= ln(2)/fz) in NCA algorithm
   - changed LSM-ref and LSM-test to LSMEAN-ref and LSMEAN-test, respectively, in the output file 
     of 'Statistical_Summaries.txt'
   - and many more that I just cannot remember right now...

v1.1.5
   - fixed the anova (lm) calculation due to the import of a .csv file. (thanks to Ji?i Hofmann, 
     Czech Republic)
   - added Type III SS (suggested by EIMaestro)
   - changed upper bound of sample size estimation from 105% to 95% (more conservative; suggested 
     by Helmut Schutz)

v1.1.4
   - fixed the compatibility for iMac and Linux (thanks to Koji Shimamoto, Tokyo, Japan; also for 
     his testing bear on iMac)

v1.1.3-1.1.0
   - removed the function of "Sample size estimation (raw data)"; also improve the function of 
     "Sample size estimation (Log Transform)."
   - fixed the rounding error in the display of "Sample size estimation (Log transform)" (thanks
     to Helmut Schutz)
   - fixed some the format (comma, semicolon, etc.) of import data file (thanks to Helmut Schutz); users 
     now can choose their favorite formats.
   - display the PATH where bear will import from and will output the all results to.
   - display only one graphic devices (using PageDown & PageUp to change plots) (thanks to Helmut Schutz)
   - display semilog (not linear) plots when choosing data points to do linear regression for 
     fz in NCA (thanks to Helmut Schutz)
   - calculate CV_intra & CV_inter now (thanks to Helmut Schutz)
   - output both the .csv and the .RData file formats obtained from NCA; users can choose either 
     one for anova.
   - use "Tests of SUBJECT(SEQUENCE) as an error term" in ANOVA output (thanks to Helmut Schutz)
   - changed ANOVA(GLM) to ANOVA (lm) in the menu title (thanks to ElMaestro)
-----

To-do lists (July 23, 2009)
-----
- add configuration setup file (as .RData data frame) into bear.  Use can create/edit this 
  configuration file. Once data has been imported, don't need to enter anything... just sit 
  back and watch. 

-----
References
1. Hauschke D, Steinijans VW, Diletti E and Burke M. Sample size determination for bioequivalence
    assessment using a multiplicative model. J. Pharmacokin. Biopharm. 20:557-561, 1992. 
2. U.S. Dept of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and
    Research.  Guidance for industry. Statistical approaches to establishing bioequivalence, 2001.
3. Liu JP and Chow SC. Sample size determination for the twp one-sided tests procedure in bioequivalence.
    J. Pharmacokin. Biopharm. 20:101-104, 1992. 
4. Schuirmann DJ. A comparison of the two one-sided tests procedure and the power approach for assessing
    the equivalence of average bioavailability. J. Pharmacokin. Biopharm. 15:657-680, 1987. 
5. Chow SC, Liu JP. Design and Analysis of Bioavailability and Bioequivalence Studies. 2009; 3rd. ed., CRC
    Press, Chapman & Hall/CRC.
6. Hauschke D, Steinijans V, Pigeot I. Bioequivalence Studies in Drug Development: Mehtods and
    Applications, 2007; John Wiley & Sons Ltd. 
