Let us start by considering parametric distributions. This uses the
-function fit_ends_mods_par(), so called because it cycles
+function fit_ends_mods(), so called because it cycles
through fitting endpoints and models. The original dataset contained
only three of these endpoints, the other three endpoints are calculated
within the function.
There is now a large difference between the STM-CR compared to the +
There is now some difference between the STM-CR compared to the STM-CF and PSM structures. All structures estimate fairly similar RMD estimates of being alive (range: 337.7 to 341.6 weeks). The estimate of mean time alive in the STM-CR structure has reduced by 52.3 weeks from 393.9 to 341.6 weeks.
The proportion of that time in the PF rather than PD state varies -considerably (PSM: 69%, STM-CF: 68.1%, STM-CR: 67.3%). Given that the -hazard of mortality would typically be lower before rather than after +somewhat (PSM: 69%, STM-CF: 68.1%, STM-CR: 67.3%). Given that the hazard +of mortality would typically be lower before rather than after progression, the pre-progression mortality is likely to be more affected by any mortality constraint that post-progression mortality. The STM-CR model of post-progression survival depends on time from progression and @@ -474,8 +474,8 @@
All structures estimate fairly similar RMD estimates of being alive (range: 334.7 to 341.6 weeks). The proportion of that time in the PF -rather than PD state varied again considerably by model structure (PSM: -69%, STM-CF: 68.1%, STM-CR: 67.3%).
+rather than PD state varied a little by model structure (PSM: 69%, +STM-CF: 68.1%, STM-CR: 67.3%).discrate optional call to calc_allrnds().Not applied here, the package also allows application of discounting
+through the discrate optional call to
+calc_allrnds(). This work could also be easily extended
+through calculating RMDs or Life Years that are Quality Adjusted
+(QALYs).
calc_drmd() and the accompanying vignette("background-mortality").create_dummydata("pharmaonc") provides a dataset based on pharmaverseadam::adsl() and pharmaverseadam::adrs_onco(). This requires a dependency to admiral and pharmaverseadam packages.compare_psm_likes(), to compare the total log-likelihood values for the patient-level dataset after fitting PSM-simple and PSM-complex models to each combination of endpoint distributions.CRAN release: 2024-05-07
+CRAN release: 2024-05-07
List containing
res: Dataset of calculation results for each model
best: Tibble indicating which is the best fitting model individually or jointly, to each endpoint, according to AIC or BIC
List containing
results: Dataset of calculation results for each model
bests: Tibble indicating which is the best fitting model individually or jointly, to each endpoint, according to AIC or BIC
# Fit parametric distributions to a dataset
bosonc <- create_dummydata("flexbosms")
parfits <- fit_ends_mods_par(bosonc)
+# \donttest{
splfits <- fit_ends_mods_spl(bosonc)
# Present comparison of likelihood calculations
-# \donttest{
compare_psm_likes(bosonc, parfits)
#> Calculating PSM simple
#> Calculating PSM complex
diff --git a/search.json b/search.json
index d4b9dcf..a546dcb 100644
--- a/search.json
+++ b/search.json
@@ -1 +1 @@
-[{"path":"https://merck.github.io/psm3mkv/LICENSE.html","id":null,"dir":"","previous_headings":"","what":"GNU General Public License","title":"GNU General Public License","text":"Version 3, 29 June 2007Copyright © 2007 Free Software Foundation, Inc. .","code":"Copyright © 2024 Merck & Co., Inc., Rahway, NJ, USA and its affiliates. All rights reserved. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program. If not, see Copyright (C) This program comes with ABSOLUTELY NO WARRANTY; for details type 'show w'. This is free software, and you are welcome to redistribute it under certain conditions; type 'show c' for details."},{"path":"https://merck.github.io/psm3mkv/articles/example.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Example use of psm3mkv","text":"vignette walks evaluating partitioned survival model (PSM) state transition model structures (either clock reset, STM-CR, clock forward types, STM-CF) dataset derived bosms3 dataset comes flexsurv package (Jackson 2016). review PSMs STMs oncology cost-effectiveness models provided Woods et al. (2020). First need load packages interest. haven’t installed psm3mkv yet, please see installation instructions install dependencies. also using Angelo Canty B. D. Ripley (2024), Xiao (2024) Wickham Henry (2023).","code":"library(\"boot\") library(\"ggsci\") library(\"psm3mkv\") library(\"purrr\")"},{"path":"https://merck.github.io/psm3mkv/articles/example.html","id":"obtaining-a-suitable-dataset","dir":"Articles","previous_headings":"","what":"Obtaining a suitable dataset","title":"Example use of psm3mkv","text":"First create suitable patient-level dataset using create_dummydata(). load data derived bosms3 dataset flexsurv package (Jackson 2016). dataset contains TTP, PFS OS data 204 patients.","code":"# Create and review the dummy dataset bosonc <- create_dummydata(\"flexbosms\") head(bosonc) #> # A tibble: 6 × 7 #> ptid pfs.durn pfs.flag os.durn os.flag ttp.durn ttp.flag #> #> 1 1 18.7 1 42.9 1 18.7 1 #> 2 2 12.0 1 23.3 1 12.0 1 #> 3 3 0.452 1 8.81 1 0.452 1 #> 4 4 9.07 1 52.7 1 9.07 1 #> 5 5 47.7 0 47.7 0 47.7 0 #> 6 6 3.26 1 13.1 1 3.26 1 summary(bosonc) #> ptid pfs.durn pfs.flag os.durn #> Min. : 1.00 Min. : 0.1071 Min. :0.0000 Min. : 0.881 #> 1st Qu.: 51.75 1st Qu.: 4.2381 1st Qu.:0.0000 1st Qu.: 7.893 #> Median :102.50 Median : 9.7262 Median :1.0000 Median :15.226 #> Mean :102.50 Mean :11.5630 Mean :0.6471 Mean :17.055 #> 3rd Qu.:153.25 3rd Qu.:16.2321 3rd Qu.:1.0000 3rd Qu.:22.729 #> Max. :204.00 Max. :48.5357 Max. :1.0000 Max. :52.702 #> os.flag ttp.durn ttp.flag #> Min. :0.0000 Min. : 0.1071 Min. :0.0000 #> 1st Qu.:0.0000 1st Qu.: 4.2381 1st Qu.:0.0000 #> Median :0.0000 Median : 9.7262 Median :1.0000 #> Mean :0.4755 Mean :11.5630 Mean :0.5049 #> 3rd Qu.:1.0000 3rd Qu.:16.2321 3rd Qu.:1.0000 #> Max. :1.0000 Max. :48.5357 Max. :1.0000"},{"path":"https://merck.github.io/psm3mkv/articles/example.html","id":"fit-survival-curves-to-the-relevant-endpoints","dir":"Articles","previous_headings":"","what":"Fit survival curves to the relevant endpoints","title":"Example use of psm3mkv","text":"three cost-effectiveness model structures considering rely modeling PFS, TTP OS, additionally three endpoints: Pre-progression death (PPD). Post progression survival function time baseline (known ‘clock forward’, PPS-CF). Post-progression survival function time progression (known ‘clock reset’, PPS-CR). suitable dataset, fit statistical models six endpoints.","code":""},{"path":"https://merck.github.io/psm3mkv/articles/example.html","id":"parametric-distributions","dir":"Articles","previous_headings":"Fit survival curves to the relevant endpoints","what":"Parametric distributions","title":"Example use of psm3mkv","text":"Let us start considering parametric distributions. uses function fit_ends_mods_par(), called cycles fitting endpoints models. original dataset contained three endpoints, three endpoints calculated within function. fitted multiple parametric distributions endpoint. need retain best-fitting distribution, select using find_bestfit_par() basis distribution lowest Akaike Information Criterion (AIC).","code":"# Create a vector of distributions of interest (flexsurv notation) alldists <- c(\"exp\", \"weibullPH\", \"llogis\", \"lnorm\", \"gamma\", \"gompertz\", \"gengamma\") # Fit all distributions to all endpoints (except gengamma to PPD and TTP) allfits_par <- fit_ends_mods_par( bosonc, cuttime = 0, ppd.dist = alldists[1:6], ttp.dist = alldists[1:6], pfs.dist = alldists, os.dist = alldists, pps_cf.dist = alldists, pps_cr.dist = alldists ) # Example 1 - PFS endpoint, distribution 2 (weibullPH) allfits_par$pfs[[2]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape 0.9313 0.8080 1.0733 0.0675 #> scale 0.0676 0.0453 0.1009 0.0138 #> #> N = 204, Events: 132, Censored: 72 #> Total time at risk: 2358.845 #> Log-likelihood = -512.0729, df = 2 #> AIC = 1028.146 # Example 2 - Parameter values for PPS-CF and PPS-CR endpoints for distribution 3 (llogis) allfits_par$pps_cf[[3]]$result$res #> est L95% U95% se #> shape 1.625037 1.264658 2.08811 0.2078833 #> scale 12.184292 8.588947 17.28465 2.1737642 allfits_par$pps_cr[[3]]$result$res #> est L95% U95% se #> shape 1.602773 1.321696 1.943626 0.1576797 #> scale 11.031674 8.769443 13.877487 1.2917274 # Pick out best distribution according to min AIC fitpar.ppd <- find_bestfit(allfits_par$ppd, \"aic\") fitpar.ttp <- find_bestfit(allfits_par$ttp, \"aic\") fitpar.pfs <- find_bestfit(allfits_par$pfs, \"aic\") fitpar.os <- find_bestfit(allfits_par$os, \"aic\") fitpar.pps_cf <- find_bestfit(allfits_par$pps_cf, \"aic\") fitpar.pps_cr <- find_bestfit(allfits_par$pps_cr, \"aic\") # Inspect the selection for PFS fitpar.pfs #> $fit #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> rate 0.05596 0.04718 0.06637 0.00487 #> #> N = 204, Events: 132, Censored: 72 #> Total time at risk: 2358.845 #> Log-likelihood = -512.5726, df = 1 #> AIC = 1027.145 #> #> #> $results #> # A tibble: 7 × 13 #> id valid conv posdef npts dists pars loglik aic bic ic rankaic #> #> 1 1 TRUE TRUE TRUE 204 exp 1 -513. 1027. 1030. 1027. 1 #> 2 2 TRUE TRUE TRUE 204 weibull… 2 -512. 1028. 1035. 1028. 3 #> 3 3 TRUE TRUE TRUE 204 llogis 2 -513. 1031. 1037. 1031. 6 #> 4 4 TRUE TRUE TRUE 204 lnorm 2 -517. 1038. 1044. 1038. 7 #> 5 5 TRUE TRUE TRUE 204 gamma 2 -512. 1028. 1035. 1028. 4 #> 6 6 TRUE TRUE TRUE 204 gompertz 2 -512. 1027. 1034. 1027. 2 #> 7 7 TRUE TRUE TRUE 204 gengamma 3 -512. 1030. 1040. 1030. 5 #> # ℹ 1 more variable: rankbic "},{"path":"https://merck.github.io/psm3mkv/articles/example.html","id":"royston-parmar-splines-models","dir":"Articles","previous_headings":"Fit survival curves to the relevant endpoints","what":"Royston-Parmar splines models","title":"Example use of psm3mkv","text":"alternative approach parametric modeling use Royston-Parmar splines (Royston Parmar 2002). can follow similar approach, using flexsurv (Jackson 2016) identify best-fitting spline distributions. six endpoints, fit 9 spline models: 1, 2 3 (internal) knots either odds, hazard normal scales. uses fit_ends_mods_spl(). fitted multiple splines endpoint. need retain best-fitting distribution, select basis distribution lowest Akaike Information Criterion (AIC). use find_bestfit() .","code":"# Fit 1-3 knot splines with all 3 scales (odds, hazard, normal) to each endpoint allfits_spl <- fit_ends_mods_spl(bosonc) # Example - PFS endpoint - 1 knot, odds scale allfits_spl$pfs[[2]]$result #> Call: #> .f(formula = ..1, k = ..2, scale = ..3) #> #> Estimates: #> est L95% U95% se #> gamma0 -2.848058 -3.298634 -2.397482 0.229890 #> gamma1 0.832337 0.449968 1.214706 0.195090 #> gamma2 -0.025453 -0.051358 0.000453 0.013217 #> #> N = 204, Events: 132, Censored: 72 #> Total time at risk: 2358.845 #> Log-likelihood = -511.7148, df = 3 #> AIC = 1029.43 allfits_spl$pfs[[2]]$result$aux$scale # Scale #> [1] \"odds\" allfits_spl$pfs[[2]]$result$aux$knots # Knot locations (log time) #> 50% #> -2.233592 2.008522 3.882300 # Pick out best distribution according to min AIC fitspl.ppd <- find_bestfit(allfits_spl$ppd, \"aic\") fitspl.ttp <- find_bestfit(allfits_spl$ttp, \"aic\") fitspl.pfs <- find_bestfit(allfits_spl$pfs, \"aic\") fitspl.os <- find_bestfit(allfits_spl$os, \"aic\") fitspl.pps_cf <- find_bestfit(allfits_spl$pps_cf, \"aic\") fitspl.pps_cr <- find_bestfit(allfits_spl$pps_cr, \"aic\") # Inspect the selection for PFS fitspl.pfs #> $fit #> Call: #> .f(formula = ..1, k = ..2, scale = ..3) #> #> Estimates: #> est L95% U95% se #> gamma0 -1.62401 -1.84681 -1.40122 0.11367 #> gamma1 0.37082 0.20155 0.54008 0.08636 #> gamma2 -0.02243 -0.03539 -0.00947 0.00661 #> #> N = 204, Events: 132, Censored: 72 #> Total time at risk: 2358.845 #> Log-likelihood = -511.4637, df = 3 #> AIC = 1028.927 #> #> #> $results #> # A tibble: 9 × 14 #> id valid conv posdef npts scales nknots pars loglik aic bic ic #> #> 1 1 TRUE TRUE TRUE 204 hazard 1 3 -512. 1030. 1040. 1030. #> 2 2 TRUE TRUE TRUE 204 odds 1 3 -512. 1029. 1039. 1029. #> 3 3 TRUE TRUE TRUE 204 normal 1 3 -511. 1029. 1039. 1029. #> 4 4 TRUE TRUE TRUE 204 hazard 2 4 -512. 1032. 1045. 1032. #> 5 5 TRUE TRUE TRUE 204 odds 2 4 -511. 1031. 1044. 1031. #> 6 6 TRUE TRUE TRUE 204 normal 2 4 -511. 1031. 1044. 1031. #> 7 7 TRUE TRUE TRUE 204 hazard 3 5 -510. 1030. 1046. 1030. #> 8 8 TRUE TRUE TRUE 204 odds 3 5 -510. 1030. 1047. 1030. #> 9 9 TRUE TRUE TRUE 204 normal 3 5 -510. 1030. 1046. 1030. #> # ℹ 2 more variables: rankaic , rankbic "},{"path":"https://merck.github.io/psm3mkv/articles/example.html","id":"combine-the-best-fits","dir":"Articles","previous_headings":"Fit survival curves to the relevant endpoints","what":"Combine the best fits","title":"Example use of psm3mkv","text":"Finally, select preferred curves endpoint. may may selected minimum AIC may parametric fits spline fits. list deliberately programmed manually - carefully. example use best fits case merely illustrates options available modeler. Let us count many parameters using model.","code":"# Bring together our preferred fits for each endpoint in a list params <- list( ppd = fitpar.ppd$fit, ttp = fitpar.ttp$fit, pfs = fitspl.pfs$fit, os = fitspl.os$fit, pps_cf = allfits_par$pps_cf[[2]]$result, pps_cr = allfits_spl$pps_cr[[2]]$result ) # Pull out number of parameters used for each endpoint count_npar <- map_vec(1:6, ~ params[[.x]]$npars) # PSM uses PFS (3) and OS (4) endpoints sum(count_npar[c(3, 4)]) #> [1] 6 # STM_CF uses PPD (1), TTP (2) and PPS_CF (5) endpoints sum(count_npar[c(1, 2, 5)]) #> [1] 6 # STM_CR uses PPD (1), TTP (2) and PPS_CR (6) endpoints sum(count_npar[c(1, 2, 6)]) #> [1] 7"},{"path":"https://merck.github.io/psm3mkv/articles/example.html","id":"comparing-likelihood-values-for-the-three-model-structures","dir":"Articles","previous_headings":"","what":"Comparing likelihood values for the three model structures","title":"Example use of psm3mkv","text":"Given selected survival modeling endpoint, can now calculate compare (log-)likelihood three model structures. can also check output ensure number parameters used model structure matches derived earlier. case, model structures fitted 203 204 patients. Among 203 patients models fitted, STM-CR model greatest likelihood (best fitting) also lowest AIC (efficient). (Since nested models, statistical distributions null hypothesis easily formed, readily derive p-value statistical significance difference.)","code":"ll_all <- calc_likes(bosonc, params) ll_all #> $all #> # A tibble: 4 × 12 #> methname npar npts_1 npts_2 npts_3 npts_4 npts_tot ll_1 ll_2 ll_3 ll_4 #> #> 1 psm_simple 7 72 29 35 68 204 -64.8 -152. -151. NA #> 2 psm_complex 8 72 29 35 68 204 -64.8 -148. -149. -471. #> 3 stm_cf 6 72 29 35 68 204 -64.7 -147. -148. -474. #> 4 stm_cr 7 72 29 35 68 204 -64.7 -147. -148. -474. #> # ℹ 1 more variable: ll_tot #> #> $valid #> # A tibble: 4 × 12 #> methname npar npts_1 npts_2 npts_3 npts_4 npts_tot ll_1 ll_2 ll_3 ll_4 #> #> 1 psm_simple 7 72 29 35 67 203 -64.8 -152. -151. -469. #> 2 psm_complex 8 72 29 35 67 203 -64.8 -148. -149. -466. #> 3 stm_cf 6 72 29 35 67 203 -64.7 -147. -148. -468. #> 4 stm_cr 7 72 29 35 67 203 -64.7 -147. -148. -468. #> # ℹ 1 more variable: ll_tot #> #> $sum #> # A tibble: 4 × 8 #> methname npts npar ll aic bic rank_aic rank_bic #> #> 1 psm_simple 203 7 -836. 1685. 1709. 4 4 #> 2 psm_complex 203 8 -828. 1672. 1698. 3 3 #> 3 stm_cf 203 6 -828. 1667. 1687. 2 1 #> 4 stm_cr 203 7 -827. 1667. 1690. 1 2"},{"path":"https://merck.github.io/psm3mkv/articles/example.html","id":"comparing-the-implied-restricted-mean-durations","dir":"Articles","previous_headings":"","what":"Comparing the implied (restricted) mean durations","title":"Example use of psm3mkv","text":"order understand degree structural uncertainty (sensitivity choice model structure), calculate (restricted) mean durations progression-free (PF) progressed disease (PD) states model type. , call calc_allrmds() function dataset statistical distributions wish consider endpoint. function also allows specification patient subset use (inclset, important bootstrapping later) time horizon. units time horizon 52.18 times shorter units output - time horizon can considered units years, whereas output units weeks. two STMs estimate duration PF state slightly longer PSM. PSM also estimates least time PD state alive overall models. STM-CF provides longest estimate time PD state overall. output can bootstrapped generate standard errors. use just 10 boostrap samples (R=10) just illustrate process. practice, want use far 10 samples. Note percentiles information reported indicates small number samples, restricted mean duration PD restricted negative PSM. indicates inconsistency statistical models used case modeling PFS OS, may additional reason STMs may preferred case.","code":"# Call the RMD functions rmd_all <- calc_allrmds(bosonc, dpam = params) # Then review the mean duration in PF, PD and total alive (OS) rmd_all$results #> # A tibble: 3 × 4 #> pf pd os model #> #> 1 22.0 5.41 27.4 PSM #> 2 20.0 10.4 30.4 STM-CF #> 3 20.0 13.3 33.3 STM-CR # Bootstrap to calculate SE over 10 bootstrap samples boot::boot( data = bosonc, statistic = calc_allrmds, R = 10, # Number of samples cuttime = 0, Ty = 10, dpam = params, boot = TRUE ) #> #> ORDINARY NONPARAMETRIC BOOTSTRAP #> #> #> Call: #> boot::boot(data = bosonc, statistic = calc_allrmds, R = 10, cuttime = 0, #> Ty = 10, dpam = params, boot = TRUE) #> #> #> Bootstrap Statistics : #> original bias std. error #> t1* 22.02965 1.6639145 3.764406 #> t2* 19.95750 1.0415112 3.079635 #> t3* 19.95750 1.0415112 3.079635 #> t4* 5.41287 -0.4906403 3.728078 #> t5* 10.41262 0.7997145 2.232518 #> t6* 13.32346 1.4157501 3.260064 #> t7* 27.44252 1.1732742 2.421223 #> t8* 30.37012 1.8412257 2.909540 #> t9* 33.28096 2.4572614 3.736715"},{"path":"https://merck.github.io/psm3mkv/articles/example.html","id":"visual-inspection-of-model-fits","dir":"Articles","previous_headings":"","what":"Visual inspection of model fits","title":"Example use of psm3mkv","text":"Creating four graphics model fit straightforward. can compare state membership probabilities PF PD states. PF curves fully overlap observed period, appear fit well visually observed PF data. big differences fit models PD membership probability. best visual fit comes PSM. STMs estimate higher probability PD membership later times observed. highest probabilities STM-CF model. Next, can look probabilities alive (.e: membership either PF PD state). , three models fit fairly well 15 weeks. closest visual fit OS curve PSM. surprising PSM involves fitting OS endpoint directly. Following PD membership graphics, STMs appear -estimate OS longer durations relative observed data. However, recall overall PSM worse fit data according likelihood, AIC BIC. Finally can look probabilities post-progression survival. observed fitted STMs PSM. STM-CR estimate follows directly fitted PPS-CR survival curve. STM-CF estimate derived based average, across patients, patients’ expected PPS-CF survival relative TTP timepoint.","code":"# Generate graphs (can take time) ptdgraphs <- graph_survs(bosonc, params) # State membership probabilities for PF state ptdgraphs$graph$pf + scale_color_npg() #> Warning: Removed 1 row containing missing values or values outside the scale range #> (`geom_line()`). # State membership probabilities for PD state ptdgraphs$graph$pd + scale_color_npg() #> Warning: Removed 1 row containing missing values or values outside the scale range #> (`geom_line()`). # State membership probabilities for OS ptdgraphs$graph$os + scale_color_npg() # Probabilities of PPS ptdgraphs$graph$pps + scale_color_npg() #> Warning: Removed 393 rows containing missing values or values outside the scale range #> (`geom_line()`)."},{"path":[]},{"path":"https://merck.github.io/psm3mkv/articles/mortality-adjustments.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Mortality adjustments","text":"second series vignettes illustrating methods evaluating fit efficiency three state oncology cost-effectiveness model structures, described accompanying journal article (Muston 2024). package heavily dependent flexsurv (Jackson 2016). fitting models, described vignette(\"example\"), estimates Restricted Mean Durations (RMDs) health states can calculated constraining background mortality given life table.","code":""},{"path":"https://merck.github.io/psm3mkv/articles/mortality-adjustments.html","id":"notation","dir":"Articles","previous_headings":"Introduction","what":"Notation","title":"Mortality adjustments","text":"denote functions survival, hazard cumulative hazard \\(S(t), h(t)\\) \\(H(t)\\). ‘unadjusted’ values relevant endpoints (e.g. PFS) dependent model structure (PSM, STM-CF STM-CR). assume ‘general background mortality’, extrapolations derivations RMD constrained, denoted subscript ‘gen’. Consequently ‘adjusted’ values, denoted superscript ‘adj’.","code":""},{"path":"https://merck.github.io/psm3mkv/articles/mortality-adjustments.html","id":"background-mortality-survival-and-hazard-functions","dir":"Articles","previous_headings":"Introduction","what":"Background mortality survival and hazard functions","title":"Mortality adjustments","text":"lifetable gives us decreasing \\(l_x\\) values points \\(t=0, ..., t_{max}\\), \\(l_{t_{max}}=0\\). Let us define background mortality survival function follows. \\[ S_{gen}(t) = 1 - R \\left(1 - \\frac{l_{x_0 + t}}{l_{x_0}} \\right) \\] \\(x_0\\) mean age study population baseline, \\(R\\) Standardized Mortality Ratio, \\(l_x\\) appropriate lifetable. can also derive average background mortality hazard \\(t\\) range \\([t_1, t_2)\\). \\[ h_{gen}(t) = \\frac{\\log[S_{gen}(t_1)] - \\log[S_{gen}(t_2)]}{t_2 - t_1} \\]","code":""},{"path":"https://merck.github.io/psm3mkv/articles/mortality-adjustments.html","id":"what-does-constraining-for-background-mortality-mean","dir":"Articles","previous_headings":"","what":"What does constraining for background mortality mean?","title":"Mortality adjustments","text":"number approaches applied literature HTA submissions, mean several things!","code":""},{"path":"https://merck.github.io/psm3mkv/articles/mortality-adjustments.html","id":"approach-1---constraining-only-the-survival-function","dir":"Articles","previous_headings":"What does constraining for background mortality mean?","what":"Approach 1 - Constraining only the survival function","title":"Mortality adjustments","text":"approach, background mortality assumed limit survival function follows. direct constraint hazard. \\[ S^{adj}(t) = \\min \\left[ S(t), S_{gen}(t) \\right] \\] simple approach, used least one NICE technology appraisal (National Institute Health Care Excellence 2022): “Patient longevity always lesser values generated disease-specific survival curve (adjustment treatment functional status) survival curve general population according age sex.” However, approach ensure hazard least great hazard background mortality.","code":""},{"path":"https://merck.github.io/psm3mkv/articles/mortality-adjustments.html","id":"approach-2---constraining-the-hazard-function","dir":"Articles","previous_headings":"What does constraining for background mortality mean?","what":"Approach 2 - Constraining the hazard function","title":"Mortality adjustments","text":"rigorous adjustment hazard background mortality acts constraint unadjusted hazard (Sweeting et al. 2023). \\[ h^{adj}(t) = \\max \\left[ h(t), h_{gen}(t) \\right] \\] \\[ \\implies S^{adj}(t) = \\exp \\left[ - \\int_0^t h^{adj}(u) du \\right] \\] either case, must assume original dataset subject background mortality. Although unlikely true, common, pragmatic assumption cost-effectiveness models long background mortality relatively insignificant trial follow-.","code":""},{"path":"https://merck.github.io/psm3mkv/articles/mortality-adjustments.html","id":"approach-3---modeling-excess-hazard-in-the-original-dataset","dir":"Articles","previous_headings":"What does constraining for background mortality mean?","what":"Approach 3 - Modeling excess hazard in the original dataset","title":"Mortality adjustments","text":"approaches work adjusting extrapolations made models fitted data assumed subject background mortality. - likely - population dataset fact subject background mortality, better model dataset using excess hazard methods rather seek make adjustments extrapolations fact. discussion survival extrapolation incorporating general population mortality provided Sweeting et al. (2023).","code":""},{"path":"https://merck.github.io/psm3mkv/articles/mortality-adjustments.html","id":"application-to-3-state-state-transition-models-stms","dir":"Articles","previous_headings":"","what":"Application to 3-state State Transition Models (STMs)","title":"Mortality adjustments","text":"discussed previously, state transition models three states may clock forward (‘STM-CF’) ‘clock reset’ (STM-CR). structures involve modeling transition directly, may performed fitting survival distributions TTP, PPD PPS endpoints. general, wish constrain pre- post-progression mortality (PPD PPS) least great background mortality. \\[ h^{adj}_{PPD}(t) = \\max \\left[h_{gen}(t), h_{PPD}(t) \\right] \\\\ h^{adj}_{PPS}(t) = \\max \\left[h_{gen}(t), h_{PPS}(t) \\right] \\\\ \\] Formulas mean time PF OS shown accompanying article respect hazard survival functions TTP, PPD PPS endpoints, PPS function time baseline Clock Forward (CF) time progression Clock Reset (CR) model. Adjusting background mortality requires using adjusted rather unadjusted hazards PPD PPS formulae. straightforward enough PPD endpoint STM-CF STM-CR structures, PPS endpoint STM-CF structure. However matters complex PPS endpoint STM-CF since hazard survival functions two times (time progression time baseline) rather just one (time baseline). package provide integral solutions relying continuous time RMDs, must instead rely discretization. Spreadsheet-based economic models almost always rely discretizing time timesteps rather follow integral formulas described continuous time. long timesteps reasonably short, restricted mean duration results remain reasonably accurate. Derided “kludge”, half-cycle corrections nevertheless recommended (Naimark, Kabboul, Krahn 2013). package provides discretized calculations.","code":""},{"path":"https://merck.github.io/psm3mkv/articles/mortality-adjustments.html","id":"application-to-partitioned-survival-models-psm","dir":"Articles","previous_headings":"","what":"Application to Partitioned Survival Models (PSM)","title":"Mortality adjustments","text":"RMD PF state time alive integrals time horizon, \\(T\\), survival functions endpoints PFS OS. Endpoints PPD PPS explicitly defined. However discussed accompanying manuscript, TTP - therefore PPD PPS endpoints directly follow PFS OS defined - often practice implicitly defined PSM economic model order calculate payoffs depend progression events. Two methods possible: Simple (default): hazard TTP relative PFS proportionate progression events observed follow-period relative total number observed PFS events. Complex: hazard TTP directly modeled fitting survival distributions endpoints. adjustment RMD estimates can proceed STMs, hazards survival functions TTP, PPD PPS reflect PSM structure (PFS OS endpoints first estimated), chosen model progression (simple complex), finally adjustment PPD PPS background mortality.","code":""},{"path":[]},{"path":"https://merck.github.io/psm3mkv/articles/mortality-adjustments.html","id":"set-up","dir":"Articles","previous_headings":"Illustration of calculations","what":"Set-up","title":"Mortality adjustments","text":"First load packages need - suggested imported psm3mkv. haven’t installed psm3mkv yet, please see installation instructions install dependencies. , making use Wickham et al. (2023), Riffe (2015) Müller Wickham (2023).","code":"library(\"dplyr\") library(\"psm3mkv\") library(\"tibble\")"},{"path":"https://merck.github.io/psm3mkv/articles/mortality-adjustments.html","id":"creating-the-life-table","dir":"Articles","previous_headings":"Illustration of calculations","what":"Creating the life table","title":"Mortality adjustments","text":"order apply constraints background mortality, need background mortality data, form lifetable. can take Human Mortality Database using HMDHFDplus package mentioned (thanks Robert Hettle recommendation). lifetable need start assumed age baseline. Lifetables constructed time years. see code run without login Human Mortality Database. Alternatively, just make mortality table. However , mortality data, can apply SMR (\\(R\\)) derive lifetable time zero required.","code":"# HMD HFD package library(\"HMDHFDplus\") # Assumed population age at baseline (time=0) baseage <- 51.0 # Mortality data - England & Wales, Female, 2019 mort <- readHMDweb( CNTRY = \"GBRTENW\", item = \"fltper_1x1\", username = \"\", password = \"\" ) |> filter(Year == 2019) |> select(Age, lx, dx) |> mutate(Timey = ceiling(Age - baseage)) |> filter(Timey >= 0) # The table needs to end with lx=0 mort <- add_row(mort, Age = 111, lx = 0, Timey = 60) mort <- tibble( Timey = 0:30, lx = 10000 * exp(-0.03 * Timey^1.1), dx = lx - lead(lx), qx = dx / lx ) # Assumed Standardized Mortality Ratio SMR <- 2 # Recalculate the lifetable with the SMR applied mort$adjlx <- mort$lx for (i in 2:length(mort$lx)) { mort$adjlx[i] <- mort$adjlx[i - 1] * (1 - SMR * mort$qx[i - 1]) } # Ensure lx>=0 mort$adjlx[mort$adjlx < 0] <- 0 # Create and view lifetable ltable <- tibble(lttime = mort$Timey, lx = mort$adjlx) head(ltable) #> # A tibble: 6 × 2 #> lttime lx #> #> 1 0 10000 #> 2 1 9409. #> 3 2 8774. #> 4 3 8151. #> 5 4 7553. #> 6 5 6985."},{"path":"https://merck.github.io/psm3mkv/articles/mortality-adjustments.html","id":"obtaining-and-fitting-distributions-to-patient-level-data","dir":"Articles","previous_headings":"Illustration of calculations","what":"Obtaining and fitting distributions to patient-level data","title":"Mortality adjustments","text":"Next get patient-level data fit PSM STMs. far, closely following vignette(\"example\"). Time assumed recorded patient-level data weeks, approximately 52.18 weeks per year.","code":"# Get some data bosonc <- create_dummydata(\"flexbosms\") # We'll make the durations a lot longer # so that they will be definitely constrained by the lifetable bosonc <- bosonc |> mutate( pfs.durn = 20 * pfs.durn, os.durn = 20 * os.durn, ttp.durn = 20 * ttp.durn ) # Fit parametric models to each endpoint allfits <- fit_ends_mods_par(bosonc) # Bring together preferred fits for each endpoint params <- list( ppd = find_bestfit(allfits$ppd, \"aic\")$fit, ttp = find_bestfit(allfits$ttp, \"aic\")$fit, pfs = find_bestfit(allfits$pfs, \"aic\")$fit, os = find_bestfit(allfits$os, \"aic\")$fit, pps_cf = find_bestfit(allfits$pps_cf, \"aic\")$fit, pps_cr = find_bestfit(allfits$pps_cr, \"aic\")$fit )"},{"path":"https://merck.github.io/psm3mkv/articles/mortality-adjustments.html","id":"making-the-projections","dir":"Articles","previous_headings":"Illustration of calculations","what":"Making the projections","title":"Mortality adjustments","text":"skipping internal external validation purposes vignette, jump instead straight making survival projections. assume 10 year time horizon. First derive projection using integral formulae, without lifetable constraints. Next derive projection using discretization approximation, still without lifetable constraints. Discretization proven fairly accurate. STM-CR estimate mean time alive reduced just 0.1 weeks example, 393.9 393.8 weeks. Finally, use discretization approximation apply lifetable constraints. now large difference STM-CR compared STM-CF PSM structures. structures estimate fairly similar RMD estimates alive (range: 337.7 341.6 weeks). estimate mean time alive STM-CR structure reduced 52.3 weeks 393.9 341.6 weeks. proportion time PF rather PD state varies considerably (PSM: 69%, STM-CF: 68.1%, STM-CR: 67.3%). Given hazard mortality typically lower rather progression, pre-progression mortality likely affected mortality constraint post-progression mortality. STM-CR model post-progression survival depends time progression therefore rather different STM-CF PSM structures, model mortality based time baseline. can check whether using ‘complex’ type PSM structure rather default ‘simple’ affects RMD estimates PSM structure. structures estimate fairly similar RMD estimates alive (range: 334.7 341.6 weeks). proportion time PF rather PD state varied considerably model structure (PSM: 69%, STM-CF: 68.1%, STM-CR: 67.3%).","code":"# Set time horizon thoz <- 10 # Run the calculations proj1 <- calc_allrmds(bosonc, dpam = params, Ty = thoz) # Present the results res1 <- proj1$results |> mutate(method = \"int\", lxadj = \"no\", psmmeth = \"simple\") res1 |> mutate( across(c(pf, pd, os), ~ num(.x, digits = 1)) ) #> # A tibble: 3 × 7 #> pf pd os model method lxadj psmmeth #> #> 1 274.4 118.3 392.7 PSM int no simple #> 2 270.2 119.7 389.9 STM-CF int no simple #> 3 270.2 123.8 393.9 STM-CR int no simple # Run the calculation proj2 <- calc_allrmds(bosonc, dpam = params, Ty = thoz, rmdmethod = \"disc\") # Present the results res2 <- proj2$results |> mutate(method = \"disc\", lxadj = \"no\", psmmeth = \"simple\") res2 |> mutate( across(c(pf, pd, os), ~ num(.x, digits = 1)) ) #> # A tibble: 3 × 7 #> pf pd os model method lxadj psmmeth #> #> 1 274.3 118.3 392.6 PSM disc no simple #> 2 270.1 119.7 389.8 STM-CF disc no simple #> 3 270.1 123.7 393.8 STM-CR disc no simple # Run the calculations proj3 <- calc_allrmds(bosonc, dpam = params, Ty = thoz, rmdmethod = \"disc\", lifetable = ltable) # Present the results res3 <- proj3$results |> mutate(method = \"disc\", lxadj = \"yes\", psmmeth = \"simple\") res3 |> mutate( across(c(pf, pd, os), ~ num(.x, digits = 1)) ) #> # A tibble: 3 × 7 #> pf pd os model method lxadj psmmeth #> #> 1 235.4 105.5 340.9 PSM disc yes simple #> 2 230.1 107.6 337.7 STM-CF disc yes simple #> 3 230.1 111.6 341.6 STM-CR disc yes simple # Proportion of time alive spent in PF proppf3 <- res3$pf / res3$os # Run the calculations proj4 <- calc_allrmds(bosonc, dpam = params, Ty = thoz, psmtype = \"complex\", rmdmethod = \"disc\", lifetable = ltable) # Present the results res4 <- proj4$results |> mutate(method = \"disc\", lxadj = \"yes\", psmmeth = \"complex\") res4 |> mutate( across(c(pf, pd, os), ~ num(.x, digits = 1)) ) #> # A tibble: 3 × 7 #> pf pd os model method lxadj psmmeth #> #> 1 231.0 103.7 334.7 PSM disc yes complex #> 2 230.1 107.6 337.7 STM-CF disc yes complex #> 3 230.1 111.6 341.6 STM-CR disc yes complex # Proportion of time alive spent in PF proppf4 <- res4$pf / res4$os"},{"path":"https://merck.github.io/psm3mkv/articles/mortality-adjustments.html","id":"comparing-the-results","dir":"Articles","previous_headings":"Illustration of calculations","what":"Comparing the results","title":"Mortality adjustments","text":"summary STM-CF estimates mean time PF state mean time alive given table . fictional case, application lifetable constraint fairly sizeable effects results - although design demonstration. summary STM-CR estimates mean time PF state mean time alive given table . application lifetable constraint similar fairly sizeable effects results also. summary PSM estimates mean time PF state mean time alive given table . change RMDs ‘complex’ PSM structure compared ‘simple’ PSM structure -4.3, -1.9 -6.2 weeks respectively time PF state, PD state time alive.","code":""},{"path":"https://merck.github.io/psm3mkv/articles/mortality-adjustments.html","id":"for-further-investigation","dir":"Articles","previous_headings":"Illustration of calculations","what":"For further investigation","title":"Mortality adjustments","text":"effect lifetable constraint STMs much greater - especially time PF state - STMs rather PSM. package also allows application discounting discrate optional call calc_allrnds().","code":""},{"path":[]},{"path":"https://merck.github.io/psm3mkv/authors.html","id":null,"dir":"","previous_headings":"","what":"Authors","title":"Authors and Citation","text":"Dominic Muston. Author, maintainer. Merck & Co., Inc., Rahway, NJ, USA affiliates. Copyright holder, funder.","code":""},{"path":"https://merck.github.io/psm3mkv/authors.html","id":"citation","dir":"","previous_headings":"","what":"Citation","title":"Authors and Citation","text":"Muston D (2024). “Informing Structural Assumptions Three State Oncology Cost-Effectiveness Models Model Efficiency Fit.” Applied Health Economics Health Policy. doi:10.1007/s40258-024-00884-2.","code":"@Article{, title = {Informing Structural Assumptions for Three State Oncology Cost-Effectiveness Models through Model Efficiency and Fit}, author = {Dominic Muston}, journal = {Applied Health Economics and Health Policy}, year = {2024}, doi = {10.1007/s40258-024-00884-2}, }"},{"path":"https://merck.github.io/psm3mkv/index.html","id":"psm3mkv-a-package-to-evaluate-the-fit-and-efficiency-of-three-state-oncology-cost-effectiveness-model-structures-","dir":"","previous_headings":"","what":"Evaluate Partitioned Survival and State Transition Models","title":"Evaluate Partitioned Survival and State Transition Models","text":"goal psm3mkv evaluate efficiency fit certain three state model structures data typical oncology clinical trial, described accompanying article (Muston 2024). package evaluates following structures: Partitioned Survival Model/analysis (PSM), Clock-forward State Transition Model (STM-CF), Clock-reset State Transition Model (STM-CR). state transition models differ transition progressive disease death function time baseline STM-CF time progression STM-CR (Jackson 2016; Woods et al. 2020). package requires patient-level dataset time progression (TTP), progression-free survival (PFS) overall survival (OS). Given , package enables: Fitting range models endpoints relevant model type: One piece parametric (distributions according flexsurv). Royston-Parmar splines (1-3 internal knots, hazard/odds/normal scales, per flexsurv) (Royston Parmar 2002). Two piece parametric (given time cutoff). Selecting “best fit” survival models endpoint (using Akaike Information Criterion, Bayesian Information Criterion user preference). Deriving presenting likelihoods 3 structures evaluate fit. Presenting total number parameters used structure, additionally evaluate efficiency. Deriving presenting restricted mean durations health state 3 model structures (given time horizon), evaluate plausibility structural sensitivity. Additional functionality: Constraining estimates ensure survival greater survival background lifetable. Applying discounting obtained discounted restricted means. Bootstrap standard errors can derived. Graphically illustrate observed fitted membership probabilities, allow visual inspection fit 3 model structures. two piece modeling used, modelers advised take care interpretation validity case different cutoff points selected different endpoints. Additionally, parametric modeling STM structures, model survival progressive disease state (post progression survival, PPS) may function additional arbitrary explanatory variable. intended enable exploration TTP (transformation) predictor PPS.","code":""},{"path":"https://merck.github.io/psm3mkv/index.html","id":"vignettes","dir":"","previous_headings":"","what":"Vignettes","title":"Evaluate Partitioned Survival and State Transition Models","text":"accompanying vignette(\"example\") illustrates package can used one-piece parametric spline modeling. second vignette, vignette(\"background-mortality\") illustrates , fitting models, estimates restricted mean durations health states can calculated constraining background mortality given life table. Survival assumed greater background lifetable.","code":""},{"path":"https://merck.github.io/psm3mkv/index.html","id":"installation","dir":"","previous_headings":"","what":"Installation","title":"Evaluate Partitioned Survival and State Transition Models","text":"package requires version R >= 4.1.0 due use native pipe. Please ensure R updated first.","code":""},{"path":"https://merck.github.io/psm3mkv/index.html","id":"latest-stable-release","dir":"","previous_headings":"Installation","what":"Latest stable release","title":"Evaluate Partitioned Survival and State Transition Models","text":"Install latest stable release CRAN:","code":"install.packages(\"psm3mkv\")"},{"path":"https://merck.github.io/psm3mkv/index.html","id":"development-version","dir":"","previous_headings":"Installation","what":"Development version","title":"Evaluate Partitioned Survival and State Transition Models","text":"Install latest development version GitHub (may stable): Note pak::pak() build vignettes default installing package GitHub, ideal vignettes can take long time generate. can conveniently view package documentation website.","code":"# install.packages(\"pak\") pak::pak(\"Merck/psm3mkv@main\")"},{"path":"https://merck.github.io/psm3mkv/index.html","id":"additional-dependencies","dir":"","previous_headings":"Installation","what":"Additional dependencies","title":"Evaluate Partitioned Survival and State Transition Models","text":"Running vignettes requires additional dependencies, either imported suggested psm3mkv. Thus can ensure installed specifying dependencies = TRUE.","code":"pak::pak(\"Merck/psm3mkv@*release\", dependencies = TRUE)"},{"path":"https://merck.github.io/psm3mkv/index.html","id":"citation","dir":"","previous_headings":"","what":"Citation","title":"Evaluate Partitioned Survival and State Transition Models","text":"use software, please cite . Muston, D. 2024. “Informing Structural Assumptions Three State Oncology Cost-Effectiveness Models Model Efficiency Fit.” Applied Health Economics Health Policy. DOI: 10.1007/s40258-024-00884-2 BibTeX entry LaTeX users ","code":"@article{muston2024informing, author = {Dominic Muston}, title = {Informing structural assumptions for three state oncology cost-effectiveness models through model efficiency and fit}, journal = {Applied Health Economics and Health Policy}, year = {2024}, doi = {10.1007/s40258-024-00884-2} }"},{"path":"https://merck.github.io/psm3mkv/index.html","id":"references","dir":"","previous_headings":"","what":"References","title":"Evaluate Partitioned Survival and State Transition Models","text":"Jackson, Christopher. 2016. “flexsurv: Platform Parametric Survival Modeling R.” Journal Statistical Software 70 (8): 1–33. Woods, Beth S, Eleftherios Sideris, Stephen Palmer, Nick Latimer, Marta Soares. 2020. “Partitioned Survival State Transition Models Healthcare Decision Making Oncology: Now?” Value Health 23 (12): 1613–1621. Royston, Patrick, Mahesh KB Parmar. 2002. “Flexible Parametric Proportional-Hazards Proportional-Odds Models Censored Survival Data, Application Prognostic Modelling Estimation Treatment Effects.” Statistics Medicine 21 (15): 2175–2197.","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_allrmds.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate restricted mean durations for each health state and all three models — calc_allrmds","title":"Calculate restricted mean durations for each health state and all three models — calc_allrmds","text":"Calculate restricted mean durations health state (progression free progressed disease) three models (partitioned survival, clock forward state transition model, clock reset state transition model).","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_allrmds.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate restricted mean durations for each health state and all three models — calc_allrmds","text":"","code":"calc_allrmds( ptdata, inclset = 0, dpam, psmtype = \"simple\", cuttime = 0, Ty = 10, lifetable = NA, discrate = 0, rmdmethod = \"int\", timestep = 1, boot = FALSE )"},{"path":"https://merck.github.io/psm3mkv/reference/calc_allrmds.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate restricted mean durations for each health state and all three models — calc_allrmds","text":"ptdata Dataset patient level data. Must tibble columns named: ptid: patient identifier pfs.durn: duration PFS baseline pfs.flag: event flag PFS (=1 progression death occurred, 0 censoring) os.durn: duration OS baseline os.flag: event flag OS (=1 death occurred, 0 censoring) ttp.durn: duration TTP baseline (usually equal pfs.durn) ttp.flag: event flag TTP (=1 progression occurred, 0 censoring). inclset Vector indicate patients include analysis dpam List statistical fits endpoint required PSM, STM-CF STM-CR models. psmtype Either \"simple\" \"complex\" PSM formulation cuttime Time cutoff - nonzero two-piece models. Ty Time duration calculate. Assumes input years, patient-level data recorded weeks. lifetable Optional, life table. Columns must include lttime (time years, 52.18 times shorter time index elsewhere, starting zero) lx discrate Discount rate (% per year) rmdmethod can \"int\" (default full integral calculations) \"disc\" approximate discretized calculations timestep required method==\"int\", default 1 boot logical flag indicate whether abbreviated output required (default = FALSE), example bootstrapping","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_allrmds.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate restricted mean durations for each health state and all three models — calc_allrmds","text":"List detailed numeric results cutadj indicates survival function area curves PFS OS cutpoint results provides results restricted means calculations, model state.","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_allrmds.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculate restricted mean durations for each health state and all three models — calc_allrmds","text":"","code":"# \\donttest{ # Create dataset and fit survival models (splines) bosonc <- create_dummydata(\"flexbosms\") fits <- fit_ends_mods_par(bosonc) # Pick out best distribution according to min AIC params <- list( ppd = find_bestfit(fits$ppd, \"aic\")$fit, ttp = find_bestfit(fits$ttp, \"aic\")$fit, pfs = find_bestfit(fits$pfs, \"aic\")$fit, os = find_bestfit(fits$os, \"aic\")$fit, pps_cf = find_bestfit(fits$pps_cf, \"aic\")$fit, pps_cr = find_bestfit(fits$pps_cr, \"aic\")$fit ) # RMD using default \"int\" method, no lifetable constraint calc_allrmds(bosonc, dpam=params) #> $cutadj #> $cutadj$pfarea #> [1] 0 #> #> $cutadj$pfsurv #> [1] 1 #> #> $cutadj$osarea #> [1] 0 #> #> $cutadj$ossurv #> [1] 1 #> #> #> $results #> # A tibble: 3 × 4 #> pf pd os model #> #> 1 17.9 9.82 27.7 PSM #> 2 20.0 10.7 30.7 STM-CF #> 3 20.0 10.1 30.0 STM-CR #> # RMD using discretized (\"disc\") method, no lifetable constraint calc_allrmds(bosonc, dpam=params, rmdmethod=\"disc\", timestep=1, boot=TRUE) #> [1] 17.87470 19.96263 19.96263 11.98304 10.73811 10.07684 29.85774 30.70075 #> [9] 30.03948 # }"},{"path":"https://merck.github.io/psm3mkv/reference/calc_haz_psm.html","id":null,"dir":"Reference","previous_headings":"","what":"Derive pre and post-progression hazards of death under PSM — calc_haz_psm","title":"Derive pre and post-progression hazards of death under PSM — calc_haz_psm","text":"Derive hazards death pre- post-progression either simple complex PSM formulations.","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_haz_psm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Derive pre and post-progression hazards of death under PSM — calc_haz_psm","text":"","code":"calc_haz_psm(timevar, ptdata, dpam, psmtype)"},{"path":"https://merck.github.io/psm3mkv/reference/calc_haz_psm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Derive pre and post-progression hazards of death under PSM — calc_haz_psm","text":"timevar Vector times calculate hazards ptdata Dataset patient level data. Must tibble columns named: ptid: patient identifier pfs.durn: duration PFS baseline pfs.flag: event flag PFS (=1 progression death occurred, 0 censoring) os.durn: duration OS baseline os.flag: event flag OS (=1 death occurred, 0 censoring) ttp.durn: duration TTP baseline (usually equal pfs.durn) ttp.flag: event flag TTP (=1 progression occurred, 0 censoring). Survival data endpoints (time progression, pre-progression death, post-progression survival) derived PFS OS. dpam List survival regressions endpoint: pre-progression death (PPD) time progression (TTP) progression-free survival (PFS) overall survival (OS) post-progression survival clock forward (PPS-CF) post-progression survival clock reset (PPS-CR). psmtype Either \"simple\" \"complex\" PSM formulation","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_haz_psm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Derive pre and post-progression hazards of death under PSM — calc_haz_psm","text":"List pre, pre-progression hazard, post, post-progression hazard","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_haz_psm.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Derive pre and post-progression hazards of death under PSM — calc_haz_psm","text":"","code":"# \\donttest{ bosonc <- create_dummydata(\"flexbosms\") fits <- fit_ends_mods_spl(bosonc) # Pick out best distribution according to min AIC params <- list( ppd = find_bestfit(fits$ppd, \"aic\")$fit, ttp = find_bestfit(fits$ttp, \"aic\")$fit, pfs = find_bestfit(fits$pfs, \"aic\")$fit, os = find_bestfit(fits$os, \"aic\")$fit, pps_cf = find_bestfit(fits$pps_cf, \"aic\")$fit, pps_cr = find_bestfit(fits$pps_cr, \"aic\")$fit ) calc_haz_psm(0:10, ptdata=bosonc, dpam=params, psmtype=\"simple\") #> $adj #> $adj$ttp #> [1] 0.00000000 0.04742078 0.04722152 0.04707233 0.04682576 0.04650723 #> [7] 0.04614387 0.04575442 0.04533051 0.04479742 0.04418546 #> #> $adj$ppd #> [1] 0.000000000 0.009903692 0.013295380 0.013253374 0.013183952 0.013094269 #> [7] 0.012991963 0.012882314 0.012762960 0.012612865 0.012440568 #> #> $adj$pfs #> [1] 0.00000000 0.05732447 0.06051690 0.06032570 0.06000971 0.05960150 #> [7] 0.05913583 0.05863674 0.05809348 0.05741028 0.05662603 #> #> $adj$os #> [1] 0.000000000 0.009300758 0.012253823 0.014430772 0.016195615 0.017678549 #> [7] 0.018943502 0.020027650 0.020955175 0.021743175 0.022404577 #> #> $adj$pps #> [1] 0.000000000 0.000000000 0.005094092 0.024838109 0.034454269 0.040318847 #> [7] 0.044430336 0.047603965 0.050244788 0.052614795 0.054815057 #> #> #> $unadj #> $unadj$ttp #> [1] 0.00000000 0.04742078 0.04722152 0.04707233 0.04682576 0.04650723 #> [7] 0.04614387 0.04575442 0.04533051 0.04479742 0.04418546 #> #> $unadj$ppd #> [1] 0.00000000 0.01335148 0.01329538 0.01325337 0.01318395 0.01309427 #> [7] 0.01299196 0.01288231 0.01276296 0.01261286 0.01244057 #> #> $unadj$pfs #> [1] 0.00000000 0.06077226 0.06051690 0.06032570 0.06000971 0.05960150 #> [7] 0.05913583 0.05863674 0.05809348 0.05741028 0.05662603 #> #> $unadj$os #> [1] 0.000000000 0.009363175 0.012472587 0.014891286 0.016981265 0.018871673 #> [7] 0.020625521 0.022278934 0.023854812 0.025368713 0.026831756 #> #> $unadj$pps #> [1] NaN 0.000000000 0.005094092 0.024838109 0.034454269 0.040318847 #> [7] 0.044430336 0.047603965 0.050244788 0.052614795 0.054815057 #> #> calc_haz_psm(0:10, ptdata=bosonc, dpam=params, psmtype=\"complex\") #> $adj #> $adj$ttp #> [1] 0.00000000 0.05758869 0.05337416 0.05075486 0.04878425 0.04717919 #> [7] 0.04581438 0.04459666 0.04337732 0.04216807 0.04098853 #> #> $adj$ppd #> [1] 0.000000000 0.003183577 0.007142746 0.009570846 0.011225462 0.012422312 #> [7] 0.013321453 0.014040074 0.014716157 0.015242215 0.015637501 #> #> $adj$pfs #> [1] 0.00000000 0.06077226 0.06051690 0.06032570 0.06000971 0.05960150 #> [7] 0.05913583 0.05863674 0.05809348 0.05741028 0.05662603 #> #> $adj$os #> [1] 0.000000000 0.009300758 0.012253823 0.014430772 0.016195615 0.017678549 #> [7] 0.018943502 0.020027650 0.020955175 0.021743175 0.022404577 #> #> $adj$pps #> [1] 0.00000000 0.11640992 0.06026855 0.04720160 0.04346609 0.04281332 #> [7] 0.04340284 0.04448366 0.04559770 0.04699859 0.04859870 #> #> #> $unadj #> $unadj$ttp #> [1] 0.00000000 0.05758869 0.05337416 0.05075486 0.04878425 0.04717919 #> [7] 0.04581438 0.04459666 0.04337732 0.04216807 0.04098853 #> #> $unadj$ppd #> [1] 0.000000000 0.003183577 0.007142746 0.009570846 0.011225462 0.012422312 #> [7] 0.013321453 0.014040074 0.014716157 0.015242215 0.015637501 #> #> $unadj$pfs #> [1] 0.00000000 0.06077226 0.06051690 0.06032570 0.06000971 0.05960150 #> [7] 0.05913583 0.05863674 0.05809348 0.05741028 0.05662603 #> #> $unadj$os #> [1] 0.000000000 0.009363175 0.012472587 0.014891286 0.016981265 0.018871673 #> [7] 0.020625521 0.022278934 0.023854812 0.025368713 0.026831756 #> #> $unadj$pps #> [1] NaN 0.11640992 0.06026855 0.04720160 0.04346609 0.04281332 #> [7] 0.04340284 0.04448366 0.04559770 0.04699859 0.04859870 #> #> # }"},{"path":"https://merck.github.io/psm3mkv/reference/calc_likes.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate likelihoods for three three-state model structures — calc_likes","title":"Calculate likelihoods for three three-state model structures — calc_likes","text":"Calculate likelihood values summary output following three state models structures: partitioned survival, clock forward state transition, clock reset state transition. function requires appropriately formatted patient-level data, set fitted survival regressions, time cut-(two-piece modeling used).","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_likes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate likelihoods for three three-state model structures — calc_likes","text":"","code":"calc_likes(ptdata, dpam, cuttime = 0)"},{"path":"https://merck.github.io/psm3mkv/reference/calc_likes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate likelihoods for three three-state model structures — calc_likes","text":"ptdata Dataset patient level data. Must tibble columns named: ptid: patient identifier pfs.durn: duration PFS baseline pfs.flag: event flag PFS (=1 progression death occurred, 0 censoring) os.durn: duration OS baseline os.flag: event flag OS (=1 death occurred, 0 censoring) ttp.durn: duration TTP baseline (usually equal pfs.durn) ttp.flag: event flag TTP (=1 progression occurred, 0 censoring). Survival data endpoints (time progression, pre-progression death, post-progression survival) derived PFS OS. dpam List survival regressions endpoint: pre-progression death (PPD) time progression (TTP) progression-free survival (PFS) overall survival (OS) post-progression survival clock forward (PPS-CF) post-progression survival clock reset (PPS-CR). cuttime Time cutoff - nonzero two-piece models.","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_likes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate likelihoods for three three-state model structures — calc_likes","text":"list three tibbles: tibble results patients: methname: model structure method. npar: number parameters used method. npts_1 npts_4 number patients experiencing outcomes 1-4 respectively (see ), npts_tot total. ll_1 ll_4 log-likelihood values patients experiencing outcomes 1-4 respectively (see ), ll_tot total. valid tibble design patients valid likelihoods 4 methods sum tibble respect patients valid likelihoods 4 methods providing: npts: number patients contributing results method. npar: number parameters used method. ll: total log-likelihood AIC: Akaike Information Criterion value model BIC: Bayesian Information Criterion value model four outcomes follows: (1) refers patients remain alive progression-free follow-; (2) refers patients die without prior progression follow-; (3) refers patients progress remain alive remaining follow-, (4) refers patients progress die within follow-.","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_likes.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculate likelihoods for three three-state model structures — calc_likes","text":"","code":"# \\donttest{ bosonc <- create_dummydata(\"flexbosms\") fits <- fit_ends_mods_spl(bosonc) # Pick out best distribution according to min AIC params <- list( ppd = find_bestfit(fits$ppd, \"aic\")$fit, ttp = find_bestfit(fits$ttp, \"aic\")$fit, pfs = find_bestfit(fits$pfs, \"aic\")$fit, os = find_bestfit(fits$os, \"aic\")$fit, pps_cf = find_bestfit(fits$pps_cf, \"aic\")$fit, pps_cr = find_bestfit(fits$pps_cr, \"aic\")$fit ) calc_likes(bosonc, dpam=params) #> $all #> # A tibble: 4 × 12 #> methname npar npts_1 npts_2 npts_3 npts_4 npts_tot ll_1 ll_2 ll_3 ll_4 #> #> 1 psm_simple 7 72 29 35 68 204 -64.8 -152. -151. NA #> 2 psm_complex 9 72 29 35 68 204 -64.8 -147. -150. -473. #> 3 stm_cf 9 72 29 35 68 204 -65.2 -147. -148. -473. #> 4 stm_cr 10 72 29 35 68 204 -65.2 -147. -148. -470. #> # ℹ 1 more variable: ll_tot #> #> $valid #> # A tibble: 4 × 12 #> methname npar npts_1 npts_2 npts_3 npts_4 npts_tot ll_1 ll_2 ll_3 ll_4 #> #> 1 psm_simple 7 72 29 35 67 203 -64.8 -152. -151. -469. #> 2 psm_complex 9 72 29 35 67 203 -64.8 -147. -150. -468. #> 3 stm_cf 9 72 29 35 67 203 -65.2 -147. -148. -468. #> 4 stm_cr 10 72 29 35 67 203 -65.2 -147. -148. -464. #> # ℹ 1 more variable: ll_tot #> #> $sum #> # A tibble: 4 × 8 #> methname npts npar ll aic bic rank_aic rank_bic #> #> 1 psm_simple 203 7 -836. 1685. 1709. 4 4 #> 2 psm_complex 203 9 -830. 1679. 1708. 3 3 #> 3 stm_cf 203 9 -828. 1673. 1703. 2 2 #> 4 stm_cr 203 10 -824. 1668. 1701. 1 1 #> # }"},{"path":"https://merck.github.io/psm3mkv/reference/calc_rmd.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate restricted mean durations — calc_rmd","title":"Calculate restricted mean durations — calc_rmd","text":"Calculates restricted mean duration, given form parametric distribution Royston-Parmar splines","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_rmd.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate restricted mean durations — calc_rmd","text":"","code":"calc_rmd(Tw, type = NA, spec = NA, survobj = NULL)"},{"path":"https://merck.github.io/psm3mkv/reference/calc_rmd.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate restricted mean durations — calc_rmd","text":"Tw time horizon (weeks) mean calculated. type either \"par\" regular parametric form (exponential, weibull etc) \"spl\" Royston-Parmar splines. spec list comprising: type==\"par\": dist statistical distribution (named per flexsurv::flexsurvreg) pars vector parameters distribution. Exponential distribution (exp) requires rate parameter. Weibull distribution (weibullPH weibull formulations) requires shape scale parameters. Log-logistic distribution (llogis) requires shape scale parameters. Log-normal distribution (lnorm) requires meanlog sdlog parameters. Gamma Gompertz distributions (gamma gompertz) require shape rate parameters. Generalized Gamma requires mu, sigma Q parameters using standard parameterization (gengamma) shape, scale k parameters using original parameterization (gengamma.orig). type==\"spl\": gamma - Vector parameters describing baseline spline function, described flexsurv::flexsurvspline. may supplied vector number elements equal length knots, case parameters common times. Alternatively matrix may supplied, rows corresponding different times, columns corresponding knots. knots - Vector locations knots axis log time, supplied increasing order. Unlike flexsurv::flexsurvspline, include two boundary knots. scale - Either \"hazard\", \"odds\", \"normal\", described flexsurv::flexsurvspline. default knots addition boundaries, model reduces Weibull, log-logistic log-normal respectively. scale must common times. survobj survival fit object flexsurv::flexsurvspline flexsurv::flexsurvreg","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_rmd.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate restricted mean durations — calc_rmd","text":"restricted mean duration, numeric value.","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_rmd.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculate restricted mean durations — calc_rmd","text":"","code":"calc_rmd(Tw=200, type=\"spl\", spec=list(gamma=c(0.1,0.2,0.1), knots=c(-5,2,4), scale=\"normal\") ) #> [1] 199.9296 calc_rmd(Tw=250, type=\"par\", spec=list(dist=\"lnorm\", pars=c(3,1)) ) #> [1] 32.45453"},{"path":"https://merck.github.io/psm3mkv/reference/calc_surv_psmpps.html","id":null,"dir":"Reference","previous_headings":"","what":"Derive PPS survival function under a PSM — calc_surv_psmpps","title":"Derive PPS survival function under a PSM — calc_surv_psmpps","text":"Derive post-progression survival (PPS) function simple complex PSM formulation.","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_surv_psmpps.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Derive PPS survival function under a PSM — calc_surv_psmpps","text":"","code":"calc_surv_psmpps(totime, fromtime = 0, ptdata, dpam, psmtype = \"simple\")"},{"path":"https://merck.github.io/psm3mkv/reference/calc_surv_psmpps.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Derive PPS survival function under a PSM — calc_surv_psmpps","text":"totime Vector times survival function calculated fromtime Vector times survival function calculated ptdata Patient-level dataset dpam List fitted survival models endpoint psmtype Either \"simple\" \"complex\" PSM formulation","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_surv_psmpps.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Derive PPS survival function under a PSM — calc_surv_psmpps","text":"Vector PPS survival function values","code":""},{"path":"https://merck.github.io/psm3mkv/reference/calc_surv_psmpps.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Derive PPS survival function under a PSM — calc_surv_psmpps","text":"","code":"# \\donttest{ bosonc <- create_dummydata(\"flexbosms\") fits <- fit_ends_mods_spl(bosonc) # Pick out best distribution according to min AIC params <- list( ppd = find_bestfit(fits$ppd, \"aic\")$fit, ttp = find_bestfit(fits$ttp, \"aic\")$fit, pfs = find_bestfit(fits$pfs, \"aic\")$fit, os = find_bestfit(fits$os, \"aic\")$fit, pps_cf = find_bestfit(fits$pps_cf, \"aic\")$fit, pps_cr = find_bestfit(fits$pps_cr, \"aic\")$fit ) calc_surv_psmpps(totime=1:10, fromtime=rep(1,10), ptdata=bosonc, dpam=params, psmtype=\"simple\") #> [1] 1.0000000 0.9995962 0.9833657 0.9541989 0.9190001 0.8807816 0.8411186 #> [8] 0.8009357 0.7607707 0.7209705 # }"},{"path":"https://merck.github.io/psm3mkv/reference/check_consistent_pfs.html","id":null,"dir":"Reference","previous_headings":"","what":"Check consistency of PFS definition Check that PFS is defined consistently with TTP and OS in a dataset. This convenience function compares pfs.durn with the lower of ttp.durn and os.durn, and checks that the event field pfs.flag is consistent with ttp.flag and os.flag (is 1 when either ttp.flag or os.flag is one). — check_consistent_pfs","title":"Check consistency of PFS definition Check that PFS is defined consistently with TTP and OS in a dataset. This convenience function compares pfs.durn with the lower of ttp.durn and os.durn, and checks that the event field pfs.flag is consistent with ttp.flag and os.flag (is 1 when either ttp.flag or os.flag is one). — check_consistent_pfs","text":"Check consistency PFS definition Check PFS defined consistently TTP OS dataset. convenience function compares pfs.durn lower ttp.durn os.durn, checks event field pfs.flag consistent ttp.flag os.flag (1 either ttp.flag os.flag one).","code":""},{"path":"https://merck.github.io/psm3mkv/reference/check_consistent_pfs.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check consistency of PFS definition Check that PFS is defined consistently with TTP and OS in a dataset. This convenience function compares pfs.durn with the lower of ttp.durn and os.durn, and checks that the event field pfs.flag is consistent with ttp.flag and os.flag (is 1 when either ttp.flag or os.flag is one). — check_consistent_pfs","text":"","code":"check_consistent_pfs(ds)"},{"path":"https://merck.github.io/psm3mkv/reference/check_consistent_pfs.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Check consistency of PFS definition Check that PFS is defined consistently with TTP and OS in a dataset. This convenience function compares pfs.durn with the lower of ttp.durn and os.durn, and checks that the event field pfs.flag is consistent with ttp.flag and os.flag (is 1 when either ttp.flag or os.flag is one). — check_consistent_pfs","text":"ds Tibble complete patient-level dataset ttp.durn, pfs.durn, os.durn durations TTP (time progression), PFS (progression-free survival), OS (overall survival). ttp.flag, pfs.flag, os.flag, pps.flag event flag indicators TTP, PFS, OS respectively (1=event, 0=censoring).","code":""},{"path":"https://merck.github.io/psm3mkv/reference/check_consistent_pfs.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check consistency of PFS definition Check that PFS is defined consistently with TTP and OS in a dataset. This convenience function compares pfs.durn with the lower of ttp.durn and os.durn, and checks that the event field pfs.flag is consistent with ttp.flag and os.flag (is 1 when either ttp.flag or os.flag is one). — check_consistent_pfs","text":"List containing: durn: Logical vector comparing expected actual PFS durations flag: Logical vector comparing expected actual PFS event flags : Single logical value TRUE durations flags match expected, FALSE otherwise","code":""},{"path":"https://merck.github.io/psm3mkv/reference/check_consistent_pfs.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Check consistency of PFS definition Check that PFS is defined consistently with TTP and OS in a dataset. This convenience function compares pfs.durn with the lower of ttp.durn and os.durn, and checks that the event field pfs.flag is consistent with ttp.flag and os.flag (is 1 when either ttp.flag or os.flag is one). — check_consistent_pfs","text":"","code":"ponc <- create_dummydata(\"pharmaonc\") check_consistent_pfs(ponc) #> $durn #> [1] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [16] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [31] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [46] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [61] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [76] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [91] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [106] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [121] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [136] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [151] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [166] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [181] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [196] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [211] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [226] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [241] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> #> $flag #> [1] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [16] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [31] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [46] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [61] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [76] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [91] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [106] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [121] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [136] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [151] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [166] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [181] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [196] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [211] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [226] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> [241] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE #> #> $all #> [1] TRUE #>"},{"path":"https://merck.github.io/psm3mkv/reference/compare_psm_likes.html","id":null,"dir":"Reference","previous_headings":"","what":"Compare likelihoods of PSMs — compare_psm_likes","title":"Compare likelihoods of PSMs — compare_psm_likes","text":"Compare total log-likelihood values patient-level dataset fitting PSM-simple PSM-complex models combination endpoint distributions","code":""},{"path":"https://merck.github.io/psm3mkv/reference/compare_psm_likes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compare likelihoods of PSMs — compare_psm_likes","text":"","code":"compare_psm_likes(ptdata, fitslist, cuttime = 0)"},{"path":"https://merck.github.io/psm3mkv/reference/compare_psm_likes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compare likelihoods of PSMs — compare_psm_likes","text":"ptdata Dataset patient level data. Must tibble columns named: ptid: patient identifier pfs.durn: duration PFS baseline pfs.flag: event flag PFS (=1 progression death occurred, 0 censoring) os.durn: duration OS baseline os.flag: event flag OS (=1 death occurred, 0 censoring) ttp.durn: duration TTP baseline (usually equal pfs.durn) ttp.flag: event flag TTP (=1 progression occurred, 0 censoring). fitslist List distribution fits relevant endpoints, calling fit_ends_mods_par() fit_ends_mods_spl() cuttime Time cutoff - nonzero two-piece models.","code":""},{"path":"https://merck.github.io/psm3mkv/reference/compare_psm_likes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compare likelihoods of PSMs — compare_psm_likes","text":"List containing res: Dataset calculation results model best: Tibble indicating best fitting model individually jointly, endpoint, according AIC BIC","code":""},{"path":"https://merck.github.io/psm3mkv/reference/compare_psm_likes.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Compare likelihoods of PSMs — compare_psm_likes","text":"","code":"# Fit parametric distributions to a dataset bosonc <- create_dummydata(\"flexbosms\") parfits <- fit_ends_mods_par(bosonc) splfits <- fit_ends_mods_spl(bosonc) # Present comparison of likelihood calculations # \\donttest{ compare_psm_likes(bosonc, parfits) #> Calculating PSM simple #> Calculating PSM complex #> Wrapping up #> $results #> # A tibble: 252 × 13 #> id ttp_meth pfs_dist os_dist ll npar npts aic bic rank_aic #> #> 1 1 simple exp exp -859. 3 204 1725. 1735. 139 #> 2 2 simple exp weibullPH -834. 4 203 1676. 1689. 38 #> 3 3 simple exp llogis -839. 4 203 1687. 1700. 93 #> 4 4 simple exp lnorm -839. 4 203 1686. 1699. 91 #> 5 5 simple exp gamma -835. 4 203 1677. 1690. 42 #> 6 6 simple exp gompertz -830. 4 203 1668. 1682. 24 #> 7 7 simple weibullPH exp -851. 4 204 1711. 1724. 126 #> 8 8 simple weibullPH weibullPH -835. 5 203 1680. 1696. 53 #> 9 9 simple weibullPH llogis -841. 5 203 1692. 1709. 108 #> 10 10 simple weibullPH lnorm -839. 5 203 1687. 1704. 95 #> # ℹ 242 more rows #> # ℹ 3 more variables: rank_bic , best_aic , best_bic #> #> $bests #> # A tibble: 4 × 5 #> ttp_meth pfs_dist os_dist meth ic #> #> 1 gompertz exp weibullPH ind aic #> 2 gompertz exp weibullPH ind bic #> 3 gompertz lnorm gamma joint aic #> 4 gompertz lnorm gamma joint bic #> compare_psm_likes(bosonc, splfits) #> Calculating PSM simple #> Calculating PSM complex #> Wrapping up #> $results #> # A tibble: 810 × 16 #> id ttp_meth ttp_knots pfs_scales pfs_knots os_scales os_knots ll npar #> #> 1 1 simple 0 hazard 3 hazard 3 -835. 7 #> 2 2 simple 0 hazard 3 odds 3 -840. 7 #> 3 3 simple 0 hazard 3 normal 3 -838. 7 #> 4 4 simple 0 hazard 3 hazard 4 -833. 8 #> 5 5 simple 0 hazard 3 odds 4 -833. 8 #> 6 6 simple 0 hazard 3 normal 4 -832. 8 #> 7 7 simple 0 hazard 3 hazard 5 -833. 9 #> 8 8 simple 0 hazard 3 odds 5 -832. 9 #> 9 9 simple 0 hazard 3 normal 5 -832. 9 #> 10 10 simple 0 odds 3 hazard 3 -825. 7 #> # ℹ 800 more rows #> # ℹ 7 more variables: npts , aic , bic , rank_aic , #> # rank_bic , best_aic , best_bic #> #> $bests #> # A tibble: 4 × 8 #> ttp_meth ttp_knots pfs_scales pfs_knots os_scales os_knots meth ic #> #> 1 normal 3 normal 3 hazard 3 ind aic #> 2 normal 3 normal 3 hazard 3 ind bic #> 3 odds 3 normal 4 hazard 4 joint aic #> 4 simple 0 odds 3 normal 4 joint bic #> # }"},{"path":"https://merck.github.io/psm3mkv/reference/constrain_survprob.html","id":null,"dir":"Reference","previous_headings":"","what":"Constrain survival probabilities according to hazards in a lifetable Recalculated constrained survival probabilities (by week) as the lower of the original unadjusted survival probability and the survival implied by the given lifetable (assumed indexed as years). — constrain_survprob","title":"Constrain survival probabilities according to hazards in a lifetable Recalculated constrained survival probabilities (by week) as the lower of the original unadjusted survival probability and the survival implied by the given lifetable (assumed indexed as years). — constrain_survprob","text":"Constrain survival probabilities according hazards lifetable Recalculated constrained survival probabilities (week) lower original unadjusted survival probability survival implied given lifetable (assumed indexed years).","code":""},{"path":"https://merck.github.io/psm3mkv/reference/constrain_survprob.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Constrain survival probabilities according to hazards in a lifetable Recalculated constrained survival probabilities (by week) as the lower of the original unadjusted survival probability and the survival implied by the given lifetable (assumed indexed as years). — constrain_survprob","text":"","code":"constrain_survprob( survprob1, survprob2 = NA, lifetable = NA, timevec = 0:(length(survprob1) - 1) )"},{"path":"https://merck.github.io/psm3mkv/reference/constrain_survprob.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Constrain survival probabilities according to hazards in a lifetable Recalculated constrained survival probabilities (by week) as the lower of the original unadjusted survival probability and the survival implied by the given lifetable (assumed indexed as years). — constrain_survprob","text":"survprob1 (Unconstrained) survival probability value vector survprob2 Optional survival probability value vector constrain (default = NA) lifetable Lifetable (default = NA) timevec Vector times corresponding survival probabilities ","code":""},{"path":"https://merck.github.io/psm3mkv/reference/constrain_survprob.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Constrain survival probabilities according to hazards in a lifetable Recalculated constrained survival probabilities (by week) as the lower of the original unadjusted survival probability and the survival implied by the given lifetable (assumed indexed as years). — constrain_survprob","text":"Vector constrained survival probabilities","code":""},{"path":"https://merck.github.io/psm3mkv/reference/constrain_survprob.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Constrain survival probabilities according to hazards in a lifetable Recalculated constrained survival probabilities (by week) as the lower of the original unadjusted survival probability and the survival implied by the given lifetable (assumed indexed as years). — constrain_survprob","text":"","code":"ltable <- tibble::tibble(lttime=0:20, lx=c(1,0.08,0.05,0.03,0.01,rep(0,16))) survprob <- c(1,0.5,0.4,0.2,0) constrain_survprob(survprob, lifetable=ltable) #> [1] 1.0 0.5 0.4 0.2 0.0 timevec <- 100*(0:4) constrain_survprob(survprob, lifetable=ltable, timevec=timevec) #> [1] 1.00000000 0.05200139 0.01201389 0.00000000 0.00000000 survprob2 <- c(1,0.45,0.35,0.15,0) constrain_survprob(survprob, survprob2) #> [1] 1.00 0.45 0.35 0.15 0.00"},{"path":"https://merck.github.io/psm3mkv/reference/create_dummydata.html","id":null,"dir":"Reference","previous_headings":"","what":"Create dummy dataset for illustration — create_dummydata","title":"Create dummy dataset for illustration — create_dummydata","text":"Create dummy dataset illustrate psm3mkv","code":""},{"path":"https://merck.github.io/psm3mkv/reference/create_dummydata.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create dummy dataset for illustration — create_dummydata","text":"","code":"create_dummydata(dsname)"},{"path":"https://merck.github.io/psm3mkv/reference/create_dummydata.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create dummy dataset for illustration — create_dummydata","text":"dsname Dataset name, follows: flexbosms provides dataset based flexsurv::bosms3(). contains fields necessary psm3mkv. Durations converted months original dataset weeks. pharmaonc provides dataset based pharmaverseadam::adsl pharmaverseadam::adrs_onco demonstrate package can used ADaM ADTTE datasets. survcan provides dataset based survival::cancer(). contains necessary ID overall survival fields . Durations converted days original dataset weeks. additionally need supply PFS TTP data (fields pfs.durn, pfs.flag, ttp.durn ttp.flag) use psm3mkv.","code":""},{"path":"https://merck.github.io/psm3mkv/reference/create_dummydata.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create dummy dataset for illustration — create_dummydata","text":"Tibble dataset, use psm3mkv functions","code":""},{"path":"https://merck.github.io/psm3mkv/reference/create_dummydata.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create dummy dataset for illustration — create_dummydata","text":"","code":"create_dummydata(\"survcan\") |> head() #> ptid os.durn os.flag #> 1 1 43.71429 1 #> 2 2 65.00000 1 #> 3 3 144.28571 0 #> 4 4 30.00000 1 #> 5 5 126.14286 1 #> 6 6 146.00000 0 create_dummydata(\"flexbosms\") |> head() #> # A tibble: 6 × 7 #> ptid pfs.durn pfs.flag os.durn os.flag ttp.durn ttp.flag #> #> 1 1 18.7 1 42.9 1 18.7 1 #> 2 2 12.0 1 23.3 1 12.0 1 #> 3 3 0.452 1 8.81 1 0.452 1 #> 4 4 9.07 1 52.7 1 9.07 1 #> 5 5 47.7 0 47.7 0 47.7 0 #> 6 6 3.26 1 13.1 1 3.26 1 create_dummydata(\"pharmaonc\") |> head() #> # A tibble: 6 × 7 #> ptid ttp.durn ttp.flag pfs.durn pfs.flag os.durn os.flag #> #> 1 01-701-1015 25.9 0 25.9 0 25.9 0 #> 2 01-701-1023 4 0 4 0 4 0 #> 3 01-701-1028 6 1 6 1 25.6 0 #> 4 01-701-1033 3.86 0 3.86 0 3.86 0 #> 5 01-701-1034 26 0 26 0 26 0 #> 6 01-701-1047 7.71 0 7.71 0 7.71 0"},{"path":"https://merck.github.io/psm3mkv/reference/create_extrafields.html","id":null,"dir":"Reference","previous_headings":"","what":"Create the additional time-to-event endpoints, adjusting for cutpoint — create_extrafields","title":"Create the additional time-to-event endpoints, adjusting for cutpoint — create_extrafields","text":"Create additional time--event endpoints, adjusting cutpoint","code":""},{"path":"https://merck.github.io/psm3mkv/reference/create_extrafields.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create the additional time-to-event endpoints, adjusting for cutpoint — create_extrafields","text":"","code":"create_extrafields(ds, cuttime = 0)"},{"path":"https://merck.github.io/psm3mkv/reference/create_extrafields.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create the additional time-to-event endpoints, adjusting for cutpoint — create_extrafields","text":"ds Patient-level dataset cuttime Time cutpoint","code":""},{"path":"https://merck.github.io/psm3mkv/reference/create_extrafields.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create the additional time-to-event endpoints, adjusting for cutpoint — create_extrafields","text":"Tibble complete patient-level dataset, adjusted cutpoint ttp.durn, pfs.durn, ppd.durn os.durn durations TTP (time progression), PFS (progression-free survival), PPD (pre-progression death) OS (overall survival) respectively beyond cutpoint. pps.durn duration survival beyond progression, irrespective cutpoint. pps.odurn difference ttp.durn os.durn (may different pps.durn). ttp.flag, pfs.flag, ppd.flag, os.flag, pps.flag event flag indicators TTP, PFS, PPD, OS PPS respectively (1=event, 0=censoring).","code":""},{"path":"https://merck.github.io/psm3mkv/reference/create_extrafields.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create the additional time-to-event endpoints, adjusting for cutpoint — create_extrafields","text":"","code":"bosonc <- create_dummydata(\"flexbosms\") create_extrafields(bosonc, cuttime=10) #> # A tibble: 204 × 16 #> ptid pfs.odurn pfs.flag os.odurn os.flag ttp.odurn ttp.flag ttp.durn #> #> 1 1 18.7 1 42.9 1 18.7 1 8.67 #> 2 2 12.0 1 23.3 1 12.0 1 2.01 #> 3 3 0.452 1 8.81 1 0.452 1 0 #> 4 4 9.07 1 52.7 1 9.07 1 0 #> 5 5 47.7 0 47.7 0 47.7 0 37.7 #> 6 6 3.26 1 13.1 1 3.26 1 0 #> 7 7 0.738 1 2.74 1 0.738 1 0 #> 8 8 35.5 1 35.5 1 35.5 0 25.5 #> 9 9 10.4 1 37.1 1 10.4 1 0.369 #> 10 10 10.5 1 32.7 1 10.5 1 0.464 #> # ℹ 194 more rows #> # ℹ 8 more variables: pfs.durn , os.durn , tzero , #> # pps.odurn , pps.durn , pps.flag , ppd.durn , #> # ppd.flag "},{"path":"https://merck.github.io/psm3mkv/reference/find_bestfit.html","id":null,"dir":"Reference","previous_headings":"","what":"Find the ","title":"Find the ","text":"multiple survival regressions fitted endpoint dataset, necessary identify preferred model. function reviews fitted regressions selects minimum Akaike Bayesian Information Criterion (AIC, BIC), depending user choice. Model fits must parametric splines.","code":""},{"path":"https://merck.github.io/psm3mkv/reference/find_bestfit.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Find the ","text":"","code":"find_bestfit(reglist, crit)"},{"path":"https://merck.github.io/psm3mkv/reference/find_bestfit.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Find the ","text":"reglist List fitted survival regressions endpoint dataset. crit Criterion used selection best fit, either \"aic\" (Akaike Information Criterion) \"bic\" (Bayesian Information Criterion).","code":""},{"path":"https://merck.github.io/psm3mkv/reference/find_bestfit.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Find the ","text":"List single survival regression best fit.","code":""},{"path":"https://merck.github.io/psm3mkv/reference/find_bestfit.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Find the ","text":"","code":"bosonc <- create_dummydata(\"flexbosms\") # Parametric modeling fits_par <- fit_ends_mods_par(bosonc) find_bestfit(fits_par$ttp, \"aic\") #> $fit #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape -0.0376 -0.0652 -0.0101 0.0141 #> rate 0.0603 0.0455 0.0800 0.0087 #> #> N = 204, Events: 103, Censored: 101 #> Total time at risk: 2358.845 #> Log-likelihood = -421.3123, df = 2 #> AIC = 846.6246 #> #> #> $results #> # A tibble: 6 × 13 #> id valid conv posdef npts dists pars loglik aic bic ic rankaic #> #> 1 1 TRUE TRUE TRUE 204 exp 1 -426. 853. 856. 853. 6 #> 2 2 TRUE TRUE TRUE 204 weibull… 2 -423. 850. 857. 850. 3 #> 3 3 TRUE TRUE TRUE 204 llogis 2 -422. 848. 855. 848. 2 #> 4 4 TRUE TRUE TRUE 204 lnorm 2 -424. 851. 858. 851. 5 #> 5 5 TRUE TRUE TRUE 204 gamma 2 -424. 851. 858. 851. 4 #> 6 6 TRUE TRUE TRUE 204 gompertz 2 -421. 847. 853. 847. 1 #> # ℹ 1 more variable: rankbic #> # \\donttest{ # Splines modeling fits_spl <- fit_ends_mods_spl(bosonc) find_bestfit(fits_spl$ttp, \"bic\") #> $fit #> Call: #> .f(formula = ..1, k = ..2, scale = ..3) #> #> Estimates: #> est L95% U95% se #> gamma0 -1.60314 -1.83224 -1.37404 0.11689 #> gamma1 0.38216 0.20690 0.55743 0.08942 #> gamma2 -0.01936 -0.03709 -0.00163 0.00905 #> #> N = 204, Events: 103, Censored: 101 #> Total time at risk: 2358.845 #> Log-likelihood = -421.5655, df = 3 #> AIC = 849.1309 #> #> #> $results #> # A tibble: 9 × 14 #> id valid conv posdef npts scales nknots pars loglik aic bic ic #> #> 1 1 TRUE TRUE TRUE 204 hazard 1 3 -423. 852. 862. 862. #> 2 2 TRUE TRUE TRUE 204 odds 1 3 -422. 849. 859. 859. #> 3 3 TRUE TRUE TRUE 204 normal 1 3 -422. 849. 859. 859. #> 4 4 TRUE TRUE TRUE 204 hazard 2 4 -423. 854. 867. 867. #> 5 5 TRUE TRUE TRUE 204 odds 2 4 -421. 851. 864. 864. #> 6 6 TRUE TRUE TRUE 204 normal 2 4 -421. 851. 864. 864. #> 7 7 TRUE TRUE TRUE 204 hazard 3 5 -420. 850. 867. 867. #> 8 8 TRUE TRUE TRUE 204 odds 3 5 -420. 850. 867. 867. #> 9 9 TRUE TRUE TRUE 204 normal 3 5 -420. 850. 867. 867. #> # ℹ 2 more variables: rankaic , rankbic #> # }"},{"path":"https://merck.github.io/psm3mkv/reference/fit_ends_mods_par.html","id":null,"dir":"Reference","previous_headings":"","what":"Fit multiple parametric survival regressions to the multiple required endpoints — fit_ends_mods_par","title":"Fit multiple parametric survival regressions to the multiple required endpoints — fit_ends_mods_par","text":"Fits multiple parametric survival regressions, according distributions stipulated, multiple endpoints required fitting partitioned survival analysis, clock forward clock reset semi-markov models.","code":""},{"path":"https://merck.github.io/psm3mkv/reference/fit_ends_mods_par.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fit multiple parametric survival regressions to the multiple required endpoints — fit_ends_mods_par","text":"","code":"fit_ends_mods_par( simdat, cuttime = 0, ppd.dist = c(\"exp\", \"weibullPH\", \"llogis\", \"lnorm\", \"gamma\", \"gompertz\"), ttp.dist = c(\"exp\", \"weibullPH\", \"llogis\", \"lnorm\", \"gamma\", \"gompertz\"), pfs.dist = c(\"exp\", \"weibullPH\", \"llogis\", \"lnorm\", \"gamma\", \"gompertz\"), os.dist = c(\"exp\", \"weibullPH\", \"llogis\", \"lnorm\", \"gamma\", \"gompertz\"), pps_cf.dist = c(\"exp\", \"weibullPH\", \"llogis\", \"lnorm\", \"gamma\", \"gompertz\"), pps_cr.dist = c(\"exp\", \"weibullPH\", \"llogis\", \"lnorm\", \"gamma\", \"gompertz\"), expvar = NA )"},{"path":"https://merck.github.io/psm3mkv/reference/fit_ends_mods_par.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Fit multiple parametric survival regressions to the multiple required endpoints — fit_ends_mods_par","text":"simdat Dataset patient level data. Must tibble columns named: ptid: patient identifier pfs.durn: duration PFS baseline pfs.flag: event flag PFS (=1 progression death occurred, 0 censoring) os.durn: duration OS baseline os.flag: event flag OS (=1 death occurred, 0 censoring) ttp.durn: duration TTP baseline (usually equal pfs.durn) ttp.flag: event flag TTP (=1 progression occurred, 0 censoring). Survival data endpoints (time progression, pre-progression death, post-progression survival) derived PFS OS. cuttime Cut-time two-piece model, equals zero one-piece models. ppd.dist Vector distributions (named per flexsurv::flexsurvreg()) fitted Pre-Progression Death (PPD). ttp.dist Vector distributions (named per flexsurv::flexsurvreg()) fitted Time Progression (TTP). pfs.dist Vector distributions (named per flexsurv::flexsurvreg()) fitted Progression-Free Survival (PFS). os.dist Vector distributions (named per flexsurv::flexsurvreg()) fitted Overall Survival (OS). pps_cf.dist Vector distributions (named per flexsurv::flexsurvreg()) fitted Post Progression Survival, time baseline (clock forward). pps_cr.dist Vector distributions (named per flexsurv::flexsurvreg()) fitted Post Progression Survival, time progression (clock reset). expvar Explanatory variable modeling PPS","code":""},{"path":"https://merck.github.io/psm3mkv/reference/fit_ends_mods_par.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fit multiple parametric survival regressions to the multiple required endpoints — fit_ends_mods_par","text":"list endpoint, distribution, containing two components: result: list class flexsurvreg containing information fitted model. error: error message returned fitting regression (NULL indicates error).","code":""},{"path":[]},{"path":"https://merck.github.io/psm3mkv/reference/fit_ends_mods_par.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Fit multiple parametric survival regressions to the multiple required endpoints — fit_ends_mods_par","text":"","code":"bosonc <- create_dummydata(\"flexbosms\") fit_ends_mods_par(bosonc, expvar=bosonc$ttp.durn) #> $ttp #> $ttp[[1]] #> $ttp[[1]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> rate 0.0437 0.0360 0.0530 0.0043 #> #> N = 204, Events: 103, Censored: 101 #> Total time at risk: 2358.845 #> Log-likelihood = -425.5134, df = 1 #> AIC = 853.0269 #> #> #> $ttp[[1]]$error #> NULL #> #> #> $ttp[[2]] #> $ttp[[2]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape 0.8379 0.7110 0.9875 0.0702 #> scale 0.0680 0.0448 0.1033 0.0145 #> #> N = 204, Events: 103, Censored: 101 #> Total time at risk: 2358.845 #> Log-likelihood = -423.0927, df = 2 #> AIC = 850.1853 #> #> #> $ttp[[2]]$error #> NULL #> #> #> $ttp[[3]] #> $ttp[[3]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape 1.0300 0.8726 1.2158 0.0872 #> scale 15.4175 11.8479 20.0625 2.0716 #> #> N = 204, Events: 103, Censored: 101 #> Total time at risk: 2358.845 #> Log-likelihood = -422.068, df = 2 #> AIC = 848.136 #> #> #> $ttp[[3]]$error #> NULL #> #> #> $ttp[[4]] #> $ttp[[4]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> meanlog 2.762 2.462 3.062 0.153 #> sdlog 1.777 1.535 2.057 0.133 #> #> N = 204, Events: 103, Censored: 101 #> Total time at risk: 2358.845 #> Log-likelihood = -423.7432, df = 2 #> AIC = 851.4864 #> #> #> $ttp[[4]]$error #> NULL #> #> #> $ttp[[5]] #> $ttp[[5]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape 0.80871 0.65294 1.00164 0.08828 #> rate 0.03166 0.02120 0.04730 0.00649 #> #> N = 204, Events: 103, Censored: 101 #> Total time at risk: 2358.845 #> Log-likelihood = -423.5205, df = 2 #> AIC = 851.0411 #> #> #> $ttp[[5]]$error #> NULL #> #> #> $ttp[[6]] #> $ttp[[6]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape -0.0376 -0.0652 -0.0101 0.0141 #> rate 0.0603 0.0455 0.0800 0.0087 #> #> N = 204, Events: 103, Censored: 101 #> Total time at risk: 2358.845 #> Log-likelihood = -421.3123, df = 2 #> AIC = 846.6246 #> #> #> $ttp[[6]]$error #> NULL #> #> #> #> $ppd #> $ppd[[1]] #> $ppd[[1]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> rate 0.01229 0.00854 0.01769 0.00228 #> #> N = 204, Events: 29, Censored: 175 #> Total time at risk: 2358.845 #> Log-likelihood = -156.5603, df = 1 #> AIC = 315.1206 #> #> #> $ppd[[1]]$error #> NULL #> #> #> $ppd[[2]] #> $ppd[[2]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape 1.43500 1.09372 1.88278 0.19884 #> scale 0.00352 0.00105 0.01183 0.00218 #> #> N = 204, Events: 29, Censored: 175 #> Total time at risk: 2358.845 #> Log-likelihood = -153.7336, df = 2 #> AIC = 311.4673 #> #> #> $ppd[[2]]$error #> NULL #> #> #> $ppd[[3]] #> $ppd[[3]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape 1.553 1.174 2.055 0.222 #> scale 43.036 29.365 63.072 8.393 #> #> N = 204, Events: 29, Censored: 175 #> Total time at risk: 2358.845 #> Log-likelihood = -154.0749, df = 2 #> AIC = 312.1498 #> #> #> $ppd[[3]]$error #> NULL #> #> #> $ppd[[4]] #> $ppd[[4]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> meanlog 3.923 3.450 4.397 0.242 #> sdlog 1.289 0.993 1.674 0.172 #> #> N = 204, Events: 29, Censored: 175 #> Total time at risk: 2358.845 #> Log-likelihood = -154.541, df = 2 #> AIC = 313.0821 #> #> #> $ppd[[4]]$error #> NULL #> #> #> $ppd[[5]] #> $ppd[[5]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape 1.5585 1.0961 2.2159 0.2799 #> rate 0.0303 0.0152 0.0605 0.0107 #> #> N = 204, Events: 29, Censored: 175 #> Total time at risk: 2358.845 #> Log-likelihood = -153.7877, df = 2 #> AIC = 311.5754 #> #> #> $ppd[[5]]$error #> NULL #> #> #> $ppd[[6]] #> $ppd[[6]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape 0.03933 0.00680 0.07185 0.01659 #> rate 0.00777 0.00436 0.01385 0.00229 #> #> N = 204, Events: 29, Censored: 175 #> Total time at risk: 2358.845 #> Log-likelihood = -154.1277, df = 2 #> AIC = 312.2554 #> #> #> $ppd[[6]]$error #> NULL #> #> #> #> $pfs #> $pfs[[1]] #> $pfs[[1]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> rate 0.05596 0.04718 0.06637 0.00487 #> #> N = 204, Events: 132, Censored: 72 #> Total time at risk: 2358.845 #> Log-likelihood = -512.5726, df = 1 #> AIC = 1027.145 #> #> #> $pfs[[1]]$error #> NULL #> #> #> $pfs[[2]] #> $pfs[[2]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape 0.9313 0.8080 1.0733 0.0675 #> scale 0.0676 0.0453 0.1009 0.0138 #> #> N = 204, Events: 132, Censored: 72 #> Total time at risk: 2358.845 #> Log-likelihood = -512.0729, df = 2 #> AIC = 1028.146 #> #> #> $pfs[[2]]$error #> NULL #> #> #> $pfs[[3]] #> $pfs[[3]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape 1.1921 1.0310 1.3784 0.0883 #> scale 11.4917 9.3244 14.1628 1.2254 #> #> N = 204, Events: 132, Censored: 72 #> Total time at risk: 2358.845 #> Log-likelihood = -513.4304, df = 2 #> AIC = 1030.861 #> #> #> $pfs[[3]]$error #> NULL #> #> #> $pfs[[4]] #> $pfs[[4]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> meanlog 2.412 2.179 2.645 0.119 #> sdlog 1.535 1.353 1.742 0.099 #> #> N = 204, Events: 132, Censored: 72 #> Total time at risk: 2358.845 #> Log-likelihood = -516.8304, df = 2 #> AIC = 1037.661 #> #> #> $pfs[[4]]$error #> NULL #> #> #> $pfs[[5]] #> $pfs[[5]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape 0.91518 0.75087 1.11544 0.09240 #> rate 0.04985 0.03633 0.06840 0.00805 #> #> N = 204, Events: 132, Censored: 72 #> Total time at risk: 2358.845 #> Log-likelihood = -512.1786, df = 2 #> AIC = 1028.357 #> #> #> $pfs[[5]]$error #> NULL #> #> #> $pfs[[6]] #> $pfs[[6]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape -0.01338 -0.03435 0.00759 0.01070 #> rate 0.06342 0.04926 0.08165 0.00817 #> #> N = 204, Events: 132, Censored: 72 #> Total time at risk: 2358.845 #> Log-likelihood = -511.7442, df = 2 #> AIC = 1027.488 #> #> #> $pfs[[6]]$error #> NULL #> #> #> #> $os #> $os[[1]] #> $os[[1]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> rate 0.02788 0.02285 0.03402 0.00283 #> #> N = 204, Events: 97, Censored: 107 #> Total time at risk: 3479.143 #> Log-likelihood = -444.2435, df = 1 #> AIC = 890.4871 #> #> #> $os[[1]]$error #> NULL #> #> #> $os[[2]] #> $os[[2]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape 1.51432 1.29287 1.77369 0.12215 #> scale 0.00559 0.00254 0.01233 0.00226 #> #> N = 204, Events: 97, Censored: 107 #> Total time at risk: 3479.143 #> Log-likelihood = -433.3483, df = 2 #> AIC = 870.6965 #> #> #> $os[[2]]$error #> NULL #> #> #> $os[[3]] #> $os[[3]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape 1.814 1.541 2.136 0.151 #> scale 23.832 20.361 27.895 1.914 #> #> N = 204, Events: 97, Censored: 107 #> Total time at risk: 3479.143 #> Log-likelihood = -436.9493, df = 2 #> AIC = 877.8987 #> #> #> $os[[3]]$error #> NULL #> #> #> $os[[4]] #> $os[[4]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> meanlog 3.1627 2.9845 3.3409 0.0909 #> sdlog 1.0166 0.8806 1.1734 0.0744 #> #> N = 204, Events: 97, Censored: 107 #> Total time at risk: 3479.143 #> Log-likelihood = -439.3941, df = 2 #> AIC = 882.7881 #> #> #> $os[[4]]$error #> NULL #> #> #> $os[[5]] #> $os[[5]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape 1.7620 1.3981 2.2207 0.2080 #> rate 0.0605 0.0435 0.0842 0.0102 #> #> N = 204, Events: 97, Censored: 107 #> Total time at risk: 3479.143 #> Log-likelihood = -434.1308, df = 2 #> AIC = 872.2617 #> #> #> $os[[5]]$error #> NULL #> #> #> $os[[6]] #> $os[[6]]$result #> Call: #> .f(formula = ..1, dist = ..2) #> #> Estimates: #> est L95% U95% se #> shape 0.04179 0.02473 0.05885 0.00871 #> rate 0.01532 0.01079 0.02174 0.00274 #> #> N = 204, Events: 97, Censored: 107 #> Total time at risk: 3479.143 #> Log-likelihood = -433.9344, df = 2 #> AIC = 871.8687 #> #> #> $os[[6]]$error #> NULL #> #> #> #> $pps_cf #> $pps_cf[[1]] #> $pps_cf[[1]]$result #> NULL #> #> $pps_cf[[1]]$error #> #> #> #> $pps_cf[[2]] #> $pps_cf[[2]]$result #> NULL #> #> $pps_cf[[2]]$error #> #> #> #> $pps_cf[[3]] #> $pps_cf[[3]]$result #> NULL #> #> $pps_cf[[3]]$error #> #> #> #> $pps_cf[[4]] #> $pps_cf[[4]]$result #> NULL #> #> $pps_cf[[4]]$error #> #> #> #> $pps_cf[[5]] #> $pps_cf[[5]]$result #> NULL #> #> $pps_cf[[5]]$error #> #> #> #> $pps_cf[[6]] #> $pps_cf[[6]]$result #> NULL #> #> $pps_cf[[6]]$error #> #> #> #> #> $pps_cr #> $pps_cr[[1]] #> $pps_cr[[1]]$result #> NULL #> #> $pps_cr[[1]]$error #>