Ajout de la vignette ORR #67

AlexisCochard · 2025-04-29T15:30:59Z

Première version de la vignette ORR

DanChaltiel · 2025-04-30T09:09:56Z

Merci pour ce premier draft, on avance c'est trop cool 😁

Petite roadmap pour cette PR :

On s'inspire de la vignette de Livia : https://oncostat.github.io/grstat/articles/kaplan_meier.html
Le but est d'avoir à la fin un document clair et lisible par des débutants en stats / recherche clinique, qui connaissent R sans être des pros.
Il faut une approche problème -> solution. En première lecture ça m'a l'air pas mal :-).
Si c'est applicable, on peut montrer des checks à faire pour que l'utilisateur vérifie qu'il/elle n'a pas fait de bêtise.
Il faut que le lecteur puisse copier-coller le code pour l'adapter à son problème.
Par conséquent, il faut le code le plus concis, donc DRY. L'IA peut aider, mais il faut vraiment que le code utilisé soit 100% compris et que l'output soit vérifié +++.
De préférence, on utilise le tidyverse exclusivement. Donc pas de crochets [], pas de sapply(). Idem, l'IA peut aider mais on fait attention.
On utilise au mieux la syntaxe markdown, notamment les backticks ` (AltGr+è puis espace sur un clavier français)
Il faut utiliser le moins possible de packages externes, car ça oblige l'utilisateur à l'installer pour installer notre package, et c'est une invitation aux embêtements en tous genres. C'est un peu contraignant je sais...
Par exemple, tu utilises GenBinomApps pour faire tes intervalles de confiance, est-ce que c'est obligé ?
On pourrait calculer des CI plus simples, et juste noter que la fonction GenBinomApps::clopper.pearson.ci() permet d'autres calculs. Ou alors récupérer leur code dans une fonction interne de grstat ?
Il faut modulariser le code en créant des fonctions (toujours le DRY). Par exemple, il faudrait une fonction report_orr(), ou flextable_orr() qui transformerait ton ouput dataframe en flextable. (peut-être pas les footnotes ?)

Pour mon dernier point, les fonctions doivent être dans un fichier R séparé (par exemple orr.R) qui sera load avec le grstat.

Lpierott

Bon travail! des petit choses a changer

Lpierott · 2025-05-16T13:36:17Z

vignettes/ORR.Rmd

+# Introduction
+
+This vignette illustrates the implementation of ORR analysis in R using `flextable` package. The ORR method is a standard approach in clinical research for estimating the percentage of people in a study or treatment group who have a partial response or complete response to the treatment within a certain period of time. A partial response is a decrease in the size of a tumor or in the amount of cancer in the body, and a complete response is the disappearance of all signs of cancer in the body. In a clinical trial, measuring the ORR is one way to see how well a new treatment works. Also called objective response rate.
+


This vignette demonstrates how to perform an Objective Response Rate (ORR) analysis in R using in R using flextable package. The ORR method is a standard approach in clinical research for estimating the percentage of patients in a study (i.e., one or more treatment groups) who experience a partial or complete response to treatment within a specified period. A partial response is a decrease in the size of a tumor or in the amount of cancer in the body compare to the baseline, and a complete response is the disappearance of all signs of cancer in the body. In a clinical trial, measuring the ORR is one way to see how well a new treatment works. Also called objective response rate.

Attention, il y a une petite confusion. On n'analyse pas l'ORR avec le package flextable, on analyse l'ORR avec des fonctions R diverses et variées, et on met le résultat dans une flextable.
Flextable n'est qu'un format de sortie comme un autre, on aurait pu utiliser gt, kable, ou tinytable pour une table HTML comme flextable, ou même partir sur un autre format comme un fichier Excel.

A terme, il était question de mettre l'appel à flextable dans une fonction pour avoir une sortie standardisée. A ce moment là, le lecteur de la vignette n'utilisera même plus flextable mais directement grstat 😊

vignettes/ORR.Rmd

Lpierott · 2025-05-16T14:32:29Z

vignettes/ORR.Rmd

+         bestresponse_withinprotocole=ifelse(previous_rcresp_num==rcresp_num, 1, 0 ),
+         .by = subjid
+  )
+


Would that be possible to just have done it in one go? so that no need to do in two steps i.e:recist_2 and unconfirmed_best_response but only in one go unconfirmed_best_response

vignettes/ORR.Rmd

Lpierott · 2025-05-19T12:04:52Z

vignettes/ORR.Rmd

+```
+
+Finally, we will use the `flextable` package for a table of results
+


Display results using flextable package

Lpierott · 2025-06-03T12:17:11Z

vignettes/ORR.Rmd

+
+```{r}
+total <- length(unique(final_unconfirmed_best_response$subjid))
+CR_unconfirmed <- length(final_unconfirmed_best_response$rcresp[final_unconfirmed_best_response$rcresp == "Complete response"])


Problem avec les total count, il faut changer a ca pour tous a mon avis, please check. merci
CR_confirmed_her2_0 <- length(her2_0$RCRESP[her2_0$RCRESP == "Complete response" & !is.na(her2_0$RCRESP)])

je filtre quelques lignes avant (dans la def recist_2) dans le code uniquement ceux qui ne sont pas NA en rcresp donc est-ce que c'est quand même utile de rajouter ici le !is.na(her2_0$RCRESP) ?

Lpierott · 2025-07-01T14:58:35Z

vignettes/ORR.Rmd

+                            round(SD_unconfirmed/total*100,1),
+                            round(PD_unconfirmed/total*100,1),
+                            round(NE_unconfirmed/total*100,1))
+


That whole chunk of code can be shortened. I am pretty sure. ask chat!

Comme ca est ce que ca marche? et comme Dan a conseille faire une function de la table.

compute_response_flextable <- function(
data,
response_column = "rcresp",
subject_id_column = "subjid",
table_title = "Unconfirmed Best Response during treatment"
) {
library(flextable)
library(officer)
library(dplyr)

total <- length(unique(data[[subject_id_column]]))

responses <- c(
"Complete response",
"Partial response",
"Stable disease",
"Progressive disease",
"Not evaluable"
)
names(responses) <- c("CR", "PR", "SD", "PD", "NE")

counts <- sapply(responses, function(r)
sum(data[[response_column]] == r & !is.na(data[[response_column]]))
)

Overall_ORR <- counts["CR"] + counts["PR"]

ci_list <- lapply(c(Overall_ORR, counts), function(n)
clopper.pearson.ci(n, total, CI = "two.sided", alpha = 0.05)
)

ci_formatted <- sapply(ci_list, function(ci)
paste0("[", round(ci$Lower.limit * 100, 1), ";", round(ci$Upper.limit * 100, 1), "]")
)

percentages <- round(c(Overall_ORR, counts) / total * 100, 1)

names_display <- c(
"Overall ORR",
"Complete response (CR)",
"Partial response (PR)",
"Stable disease (SD)",
"Progressive disease (PD)",
"Non evaluable (NE)"
)

result_df <- data.frame(
Name = names_display,
N = c(Overall_ORR, counts),
Percentage = percentages,
IC 95% = ci_formatted,
stringsAsFactors = FALSE
)

colnames(result_df) <- c(
table_title,
paste0("N=", total),
"%",
"IC 95%"
)

ft <- result_df %>%
flextable() %>%
bold(i = 1, j = 1, bold = TRUE, part = "body") %>%
bold(bold = TRUE, part = "header") %>%
surround(i = 1, j = 1:4,
border.bottom = fp_border(color = "black", style = "solid", width = 1),
part = "body") %>%
footnote(
i = 1, j = 4,
value = as_paragraph(
"Clopper-Pearson (Exact) method was used for confidence interval"
),
ref_symbols = "*",
part = "header"
) %>%
footnote(
i = 6, j = 1,
value = as_paragraph(
"Example: Patient 129 had NE as her global response for target lesion was NE."
),
ref_symbols = "1",
part = "body"
) %>%
valign(valign = "bottom", part = "header")

return(ft)
}

my_flextable <- compute_response_flextable(final_unconfirmed_best_response)
my_flextable

Vous avez regarder ce que j avais essayer un peu de faire au dessus? j ai compris que du coup vous avez decide de faire une function qui calcul les ORR automatiquement depuis RECIST data. Je pensais que a la reunion grstat on avait decide avec Matthieu que c etait pas une bonne idee de faire ca car chaque protocol est different etc... et donc l'idee de d'expliquer comment calculer les ORR dans une vignette et presenter les ORR dans une table etait plus favorable. please let me know.

A notre dernière réunion avec Matthieu, on avait remarqué que la fonction waterfall-plot devait être séparée en une fonction qui calcule la best response et une qui plot le resultat.
C'est ce que j'ai fait dans #71 (issue #19).
C'est vrai qu'on a beaucoup dit qu'on ne voulait pas de fonction toute faite, mais il faut croire qu'on tend quand même vers ça...

Sauf qu'en fait, best response c'est l'ORR, je n'avais pas bien compris les termes (pour moi ORR c'était réponse à 1an par exemple).
Donc en discutant avec Alexis, on s'est rendus compte que le travail avait été fait en double, et qu'il fallait fusionner les deux.
C'est super car ça va permettre d'intégrer l'option "confirmed" aux waterfall plots!

Ca n'empêche qu'il faudra tout expliquer dans la vignette !

Lpierott · 2025-07-02T08:50:31Z

vignettes/ORR.Rmd

+
+## Data Overview
+
+To calculate the ORR, we'll use the “recist” dataset.


we will and not we'll. I think is too informal the we'll.

vignettes/ORR.Rmd

Lpierott · 2025-07-02T09:01:47Z

vignettes/ORR.Rmd

+
+
+#Since we have 2 arms, we will first creae the part of the control
+


Please see the code above and check and see if we can apply it here too

This reverts commit 5bf6a5f.

DanChaltiel

Je t'ai fait une review éclair vu qu'on se voit demain.
Je n'ai pas eu le temps de tester sur les données, j'essaierai demain matin, mais ça a l'air super :-)

Au fait attention, il y a survfit_stack.R qui s'est glissé dans ta branche, je ne sais pas pourquoi mais ça risque de faire des bugs quand on va merger

DanChaltiel · 2025-07-28T14:56:06Z

R/ORR_table.R

+#' @importFrom cli cli_abort
+
+ORR_table = function(id = recist$SUBJID, global_response = recist$RCRESP, date = recist$RCDT, confirmed = FALSE, show_CBR = FALSE){
+  `%notin%` <- Negate(`%in%`)


Il y a vraiment besoin d'un opérateur en plus ?

if(x %notin% b) ... if(! x %in% b) ...

DanChaltiel · 2025-07-28T14:58:02Z

R/ORR_table.R

+  if(length(show_CBR) > 1){
+    cli_abort("show_CBR shoulb be TRUE or FALSE (default = FALSE)")
+  }
+  if(confirmed %notin% c(TRUE,FALSE,NA)){


tu peux utiliser les fonctions du fichier assertion.R
Par exemple ici assert_class(confirmed, "numeric")

DanChaltiel · 2025-07-28T15:03:36Z

R/ORR_table.R

+#' @importFrom officer fp_border
+#' @importFrom cli cli_abort
+
+ORR_table = function(id = recist$SUBJID, global_response = recist$RCRESP, date = recist$RCDT, confirmed = FALSE, show_CBR = FALSE){


Dans grstat, on utilise des interfaces tidyverse, pas base R comme ici
ça veut dire que le premier argument est quasi toujours une dataframe
donc il vaut mieux avoir dans tes arguments d'abord data, puis un moyen d'identifier les colonnes, par exemple subjid="SUBJID", resp="RCRESP".

Attention, tes valeurs par défaut attendent qu'un objet recist existe, ce qui fera planter la fonction dans quasi tous les cas de figures réels.

DanChaltiel · 2025-07-28T15:05:22Z

R/ORR_table.R

+  #Si un patient a une seule visite recist (la premiere) puis aucune autre, on le modifie en Not evaluable
+  data = data %>%
+    distinct(subjid,rcresp,rcdt) %>%
+    arrange(as.numeric(subjid), rcdt) %>%


cette ligne plantera dans toutes les études où SUBJID n'est pas numérique

DanChaltiel · 2025-07-28T15:06:56Z

R/ORR_table.R

+    mutate(n = n(), .by = subjid) %>%
+    mutate(rcresp = ifelse(n==1, "Not evaluable",as.character(rcresp)))
+
+  data$rcresp <- factor(data$rcresp, levels = c("Complete response", "Partial response", "Stable disease", "Progressive disease", "Not evaluable"))


Au secours, du base R 😱😱

DanChaltiel · 2025-07-28T15:19:24Z

R/ORR_table.R

+      filter(bestresponse==rcresp_num) %>%
+      slice_head(by=subjid) %>%
+      mutate(Overall_ORR= ifelse(rcresp=="Complete response" | rcresp=="Partial response",1,0),
+             CBR = ifelse(duree_suivi_max >= 152 | rcresp=="Complete response" | rcresp=="Partial response",1,0))


Pourquoi 152 ?
C'est une valeur fixe pour toutes les études ?
Sinon il faudrait l'ajouter en argument non ?

DanChaltiel · 2025-07-28T15:28:38Z

R/ORR_table.R

+    total <- length(unique(final_best_response$subjid))
+    CR <- length(final_best_response$rcresp[final_best_response$rcresp == "Complete response"])
+    PR <- length(final_best_response$rcresp[final_best_response$rcresp == "Partial response"])
+    SD <- length(final_best_response$rcresp[final_best_response$rcresp == "Stable disease"])
+    PD <- length(final_best_response$rcresp[final_best_response$rcresp == "Progressive disease"])
+    NE <- length(final_best_response$rcresp[final_best_response$rcresp == "Not evaluable"])
+    Overall_ORR <- CR + PR
+    CBR <- length(final_best_response$CBR[final_best_response$CBR==1])
+
+    CR_CP <- clopper.pearson.ci(CR,total,CI="two.sided", alpha = 0.05)
+    PR_CP <- clopper.pearson.ci(PR,total,CI="two.sided", alpha = 0.05)
+    SD_CP <- clopper.pearson.ci(SD,total,CI="two.sided", alpha = 0.05)
+    PD_CP <- clopper.pearson.ci(PD,total,CI="two.sided", alpha = 0.05)
+    NE_CP <- clopper.pearson.ci(NE,total,CI="two.sided", alpha = 0.05)
+    Overall_ORR_CP <- clopper.pearson.ci(Overall_ORR,total,CI="two.sided", alpha = 0.05)
+    CBR_CP <- clopper.pearson.ci(CBR,total,CI="two.sided", alpha = 0.05)
+
+    CR_CP_IC <- paste0("[",round(CR_CP$Lower.limit*100,1),";",round(CR_CP$Upper.limit*100,1),"]")
+    PR_CP_IC <- paste0("[",round(PR_CP$Lower.limit*100,1),";",round(PR_CP$Upper.limit*100,1),"]")
+    SD_CP_IC <- paste0("[",round(SD_CP$Lower.limit*100,1),";",round(SD_CP$Upper.limit*100,1),"]")
+    PD_CP_IC <- paste0("[",round(PD_CP$Lower.limit*100,1),";",round(PD_CP$Upper.limit*100,1),"]")
+    NE_CP_IC <- paste0("[",round(NE_CP$Lower.limit*100,1),";",round(NE_CP$Upper.limit*100,1),"]")
+    Overall_ORR_CP_IC <- paste0("[",round(Overall_ORR_CP$Lower.limit*100,1),";",round(Overall_ORR_CP$Upper.limit*100,1),"]")
+    CBR_CP_IC <- paste0("[",round(CBR_CP$Lower.limit*100,1),";",round(CBR_CP$Upper.limit*100,1),"]")


Attention, beaucoup de baseR et beaucoup de répétitions (DRY), donc risque d'erreur et mauvaise maintenabilité.

final_best_response %>% mutate(total=n()) %>% count(rcresp) devrait être un début, faudrait que je teste pour le reste. On voit ensemble si besoin évidemment :-)

DanChaltiel · 2025-07-28T15:29:28Z

R/ORR_table.R

+      IC_95 <- c(Overall_ORR_CP_IC,
+                 CR_CP_IC,
+                 PR_CP_IC,
+                 SD_CP_IC,
+                 PD_CP_IC,
+                 NE_CP_IC)


Bon alors là je vais dire le contraire, c'est un peu trop aéré 😂
Si ça prend trop de lignes c'est aussi difficile à lire.

DanChaltiel · 2025-07-28T15:32:37Z

R/ORR_table.R

+
+    colnames(Best_Response_during_treatment) <- nom_col
+    Best_Response_during_treatment =  Best_Response_during_treatment %>%
+      as_flextable(show_coltype = F, include.row_percent = FALSE, include.column_percent = FALSE, include.table_percent = FALSE) %>%


Best_Response_during_treatment est une dataframe, donc à mon avis as_flextable() est plus galère que simplement flextable().

DanChaltiel · 2025-07-28T15:34:27Z

R/ORR_table.R

+    }
+
+    colnames(Best_Response_during_treatment) <- nom_col
+    Best_Response_during_treatment =  Best_Response_during_treatment %>%


Il faut toujours séparer le fond et la forme.
Ca veut dire qu'il faut toujours une fonction qui fabrique les données (dataframe) et une fonction qui les affiche (flextable).
Ca implique de faire un peu de programmation orientée objet, on voit ça ensemble.

This reverts commit 632fb01.

This reverts commit 15a2cff.

commit f220fb5 Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Fri Jul 18 14:08:20 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9015 [skip ci] commit 972e13e Author: Lpierott <pierotti.livia@gmail.com> Date: Fri Jul 18 16:06:01 2025 +0200 Improve KaplanMeier vignette (#74) * Create kaplan_meier_v2.Rmd * remove v2 to the name of the rmd script commit 6f32680 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Tue Jul 15 10:08:48 2025 +0200 Delete .github/ISSUE_TEMPLATE directory [skip ci] commit bf2d483 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Sat Jul 12 14:09:51 2025 +0200 Waterfallplot: séparation calcul réponse / plot (#71) * removed worst_resp calculation! * isolate calc_best_response() * add .remove_post_pd() * improve/fix inner checks * isolate .encode_response * use slice_min instead of filter for better tie management * output: select and rename +clean +add ellipsis argument * grstat_data_warn default subjid * doc * waterfall: change interface from raw recist to best response * add .check_legacy() error message to signal the function has changed * add .get_shape_layer() more versatile than stars * add .get_fill_scale() get fill colors from parameter "resp_colors" * clean labs and theme * doc * autoimport * add tests * signal dev lifecycle * grstat_dev_warn: add verbosity control for tests * improve globalVariables management * fixes for check * grstat_dev_warn fix * Squashed commit of the following: commit 63d0d15 Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Fri Jul 11 16:07:16 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9013 [skip ci] commit f95b324 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Fri Jul 11 18:06:07 2025 +0200 Implémenter les checks RECIST (#63) * WIP * recist_check: .add_to_recist_issues() instead of warn use an environment to centralize the issues instead of warning for each. + heavy reformat (should have made 2 commits...) * recist_check reorder * check_recist: .apply_recist_mapping() * check_lesion_number() * fixed .apply_recist_mapping * check_constancy * check_baseline_lesions * hotfix: add rtn objects * check_constancy: Response should be constant per date * check_baseline_lesions: baseline response should be missing * check_derived_columns * check_lesion_number() becomes check_target_lesions() * check_missing * check_target_response * fix check_missing * check_constancy: add nontarget_yn * check_target_response (for real) * check_global_response * .summarise_recist (for rc_short) * move utils to utils.R * check_global_response: fix arguments * .apply_recist_mapping: add mandatory variables * recist_issue: allow newline in message * backup commit * remove old code OLD_check_bare_recist OLD_check_response FIX_check_crf_structure * check_nontarget_response Patients with no Non-Target Lesions should not have a NTL response. * check_global_response: check NTL progressions Progressions due only to Non-Target Lesions are seldom and should be checked * remove legacy code * print.check_recist * remove unused grstat_env * gr_recist_mapping: remove unused variables * print.check_recist : nicer * document * check_recist: add mapping as attribute * rephrase most issue messages * recist_report() !!! * autoimport * check_derived_columns: conditional checks * improve documentation * Update _pkgdown.yml * recist_report: add title parameter * Delete test-recist.R * add globalVariables * fix for check * check_derived_columns: deduplicate checks * check_global_response: fix nlt-only progressions * template recist: improve options remove rownames, remove research * add assert_extension() * recist_report: split into HTML and XLSX * move to recist_utils.R * renamed to .recist_to_num() (and improved) * Remove rows posterior to the first progression * rewrote error messages * check that CR are terminal * check: Target Lesion should be <2 lymph nodes * signal dev state improves #69 * clean browser() reliquates * Update .gitignore * check_global_response: fix NA/NE * improve globalVariables() management * improve doc * fix for check * Update .gitignore * remove nonsensical check nadir can evolve over time, obviously * prefix internal check function with rc_ * add some tests * Update _pkgdown.yml * update doc --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> commit 48afe41 Author: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> Date: Fri Jul 11 14:05:35 2025 +0200 changed check action [skip ci] commit 33dd5ee Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Fri Jul 11 10:36:13 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9012 [skip ci] commit 68d87be Author: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> Date: Fri Jul 11 12:35:04 2025 +0200 Update check-standard.yaml commit 42db0f2 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Fri Jul 11 12:07:07 2025 +0200 add BOIN plot (#77) * add boin plot * autoimport * doc * add tests * Update _pkgdown.yml * fix for check * legend inside the gantt plot * clean unused variable * use boin$boundary_tab if full_boundary_tab is not available * improve documentation * update tests with newer example * fix for check * renamed data_patients columns + doc * test: check for unknown dose * typo * update tests * Update .gitignore * add experimental badge * update globals --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> commit ebbf206 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Wed Jul 9 15:52:32 2025 +0200 Update check-standard.yaml [skip ci] commit a65605f Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Wed Jul 9 09:13:24 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9011 [skip ci] commit 05b7e51 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Wed Jul 9 11:11:56 2025 +0200 Improve recist data (#75) * better arm repartition * improve date_inclusion calculation * improve documentation * example_rc: removed rctlsum_b & rctlmin * moved test-alexis.R to test-data.R * example_rc: add NTL yesno column `rcntlyn` NTL are NE mostly when non present * example_rc: fixed labels bind_rows removed them, added at last * improved documentation * example_rc: reformat * add .sample_yesno() * example_rc: improve workflow * example_rc: fix number of timepoints baseline was added * example_rc: removed unused variables * example_rc: minor reformat, renaming, and doc * example_rc: internalise things inside .simulate_one_patient() rctlsum_b + calculate rcdt from delai example_rc: simulate baseline inside .simulate_one_patient() also, decrease linearly (% of baseline) instead of exponentially (% of previous date) for simplicity * example_rc: revert baseline being not missing +response factor level management * example_rc: tweak default values * add tests on RECIST data (survival rates...) * add dev tests on RECIST (skipped, for viz only) * fix for check * add missing values * update doc * add NA/NE for real this time * update test for NE values --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> * Update _pkgdown.yml * Squashed commit of the following: commit a304e92 Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Fri Jul 11 16:58:20 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9014 [skip ci] commit 165566a Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Fri Jul 11 18:56:53 2025 +0200 new gr_officer_template() (#72) * new gr_officer_template() * doc * Update _pkgdown.yml * autoimport * fix for check * update globals --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> commit 63d0d15 Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Fri Jul 11 16:07:16 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9013 [skip ci] commit f95b324 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Fri Jul 11 18:06:07 2025 +0200 Implémenter les checks RECIST (#63) * WIP * recist_check: .add_to_recist_issues() instead of warn use an environment to centralize the issues instead of warning for each. + heavy reformat (should have made 2 commits...) * recist_check reorder * check_recist: .apply_recist_mapping() * check_lesion_number() * fixed .apply_recist_mapping * check_constancy * check_baseline_lesions * hotfix: add rtn objects * check_constancy: Response should be constant per date * check_baseline_lesions: baseline response should be missing * check_derived_columns * check_lesion_number() becomes check_target_lesions() * check_missing * check_target_response * fix check_missing * check_constancy: add nontarget_yn * check_target_response (for real) * check_global_response * .summarise_recist (for rc_short) * move utils to utils.R * check_global_response: fix arguments * .apply_recist_mapping: add mandatory variables * recist_issue: allow newline in message * backup commit * remove old code OLD_check_bare_recist OLD_check_response FIX_check_crf_structure * check_nontarget_response Patients with no Non-Target Lesions should not have a NTL response. * check_global_response: check NTL progressions Progressions due only to Non-Target Lesions are seldom and should be checked * remove legacy code * print.check_recist * remove unused grstat_env * gr_recist_mapping: remove unused variables * print.check_recist : nicer * document * check_recist: add mapping as attribute * rephrase most issue messages * recist_report() !!! * autoimport * check_derived_columns: conditional checks * improve documentation * Update _pkgdown.yml * recist_report: add title parameter * Delete test-recist.R * add globalVariables * fix for check * check_derived_columns: deduplicate checks * check_global_response: fix nlt-only progressions * template recist: improve options remove rownames, remove research * add assert_extension() * recist_report: split into HTML and XLSX * move to recist_utils.R * renamed to .recist_to_num() (and improved) * Remove rows posterior to the first progression * rewrote error messages * check that CR are terminal * check: Target Lesion should be <2 lymph nodes * signal dev state improves #69 * clean browser() reliquates * Update .gitignore * check_global_response: fix NA/NE * improve globalVariables() management * improve doc * fix for check * Update .gitignore * remove nonsensical check nadir can evolve over time, obviously * prefix internal check function with rc_ * add some tests * Update _pkgdown.yml * update doc --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> commit 48afe41 Author: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> Date: Fri Jul 11 14:05:35 2025 +0200 changed check action [skip ci] commit 33dd5ee Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Fri Jul 11 10:36:13 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9012 [skip ci] commit 68d87be Author: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> Date: Fri Jul 11 12:35:04 2025 +0200 Update check-standard.yaml commit 42db0f2 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Fri Jul 11 12:07:07 2025 +0200 add BOIN plot (#77) * add boin plot * autoimport * doc * add tests * Update _pkgdown.yml * fix for check * legend inside the gantt plot * clean unused variable * use boin$boundary_tab if full_boundary_tab is not available * improve documentation * update tests with newer example * fix for check * renamed data_patients columns + doc * test: check for unknown dose * typo * update tests * Update .gitignore * add experimental badge * update globals --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> commit ebbf206 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Wed Jul 9 15:52:32 2025 +0200 Update check-standard.yaml [skip ci] commit a65605f Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Wed Jul 9 09:13:24 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9011 [skip ci] commit 05b7e51 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Wed Jul 9 11:11:56 2025 +0200 Improve recist data (#75) * better arm repartition * improve date_inclusion calculation * improve documentation * example_rc: removed rctlsum_b & rctlmin * moved test-alexis.R to test-data.R * example_rc: add NTL yesno column `rcntlyn` NTL are NE mostly when non present * example_rc: fixed labels bind_rows removed them, added at last * improved documentation * example_rc: reformat * add .sample_yesno() * example_rc: improve workflow * example_rc: fix number of timepoints baseline was added * example_rc: removed unused variables * example_rc: minor reformat, renaming, and doc * example_rc: internalise things inside .simulate_one_patient() rctlsum_b + calculate rcdt from delai example_rc: simulate baseline inside .simulate_one_patient() also, decrease linearly (% of baseline) instead of exponentially (% of previous date) for simplicity * example_rc: revert baseline being not missing +response factor level management * example_rc: tweak default values * add tests on RECIST data (survival rates...) * add dev tests on RECIST (skipped, for viz only) * fix for check * add missing values * update doc * add NA/NE for real this time * update test for NE values --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> * deduplicate `.remove_post_pd()` * update globals * .check_best_resp: consider NE as NA * doc * update snapshots with new recist data * Squashed commit of the following: commit a304e92 Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Fri Jul 11 16:58:20 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9014 [skip ci] commit 165566a Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Fri Jul 11 18:56:53 2025 +0200 new gr_officer_template() (#72) * new gr_officer_template() * doc * Update _pkgdown.yml * autoimport * fix for check * update globals --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> commit 63d0d15 Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Fri Jul 11 16:07:16 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9013 [skip ci] commit f95b324 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Fri Jul 11 18:06:07 2025 +0200 Implémenter les checks RECIST (#63) * WIP * recist_check: .add_to_recist_issues() instead of warn use an environment to centralize the issues instead of warning for each. + heavy reformat (should have made 2 commits...) * recist_check reorder * check_recist: .apply_recist_mapping() * check_lesion_number() * fixed .apply_recist_mapping * check_constancy * check_baseline_lesions * hotfix: add rtn objects * check_constancy: Response should be constant per date * check_baseline_lesions: baseline response should be missing * check_derived_columns * check_lesion_number() becomes check_target_lesions() * check_missing * check_target_response * fix check_missing * check_constancy: add nontarget_yn * check_target_response (for real) * check_global_response * .summarise_recist (for rc_short) * move utils to utils.R * check_global_response: fix arguments * .apply_recist_mapping: add mandatory variables * recist_issue: allow newline in message * backup commit * remove old code OLD_check_bare_recist OLD_check_response FIX_check_crf_structure * check_nontarget_response Patients with no Non-Target Lesions should not have a NTL response. * check_global_response: check NTL progressions Progressions due only to Non-Target Lesions are seldom and should be checked * remove legacy code * print.check_recist * remove unused grstat_env * gr_recist_mapping: remove unused variables * print.check_recist : nicer * document * check_recist: add mapping as attribute * rephrase most issue messages * recist_report() !!! * autoimport * check_derived_columns: conditional checks * improve documentation * Update _pkgdown.yml * recist_report: add title parameter * Delete test-recist.R * add globalVariables * fix for check * check_derived_columns: deduplicate checks * check_global_response: fix nlt-only progressions * template recist: improve options remove rownames, remove research * add assert_extension() * recist_report: split into HTML and XLSX * move to recist_utils.R * renamed to .recist_to_num() (and improved) * Remove rows posterior to the first progression * rewrote error messages * check that CR are terminal * check: Target Lesion should be <2 lymph nodes * signal dev state improves #69 * clean browser() reliquates * Update .gitignore * check_global_response: fix NA/NE * improve globalVariables() management * improve doc * fix for check * Update .gitignore * remove nonsensical check nadir can evolve over time, obviously * prefix internal check function with rc_ * add some tests * Update _pkgdown.yml * update doc --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> commit 48afe41 Author: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> Date: Fri Jul 11 14:05:35 2025 +0200 changed check action [skip ci] commit 33dd5ee Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Fri Jul 11 10:36:13 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9012 [skip ci] commit 68d87be Author: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> Date: Fri Jul 11 12:35:04 2025 +0200 Update check-standard.yaml commit 42db0f2 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Fri Jul 11 12:07:07 2025 +0200 add BOIN plot (#77) * add boin plot * autoimport * doc * add tests * Update _pkgdown.yml * fix for check * legend inside the gantt plot * clean unused variable * use boin$boundary_tab if full_boundary_tab is not available * improve documentation * update tests with newer example * fix for check * renamed data_patients columns + doc * test: check for unknown dose * typo * update tests * Update .gitignore * add experimental badge * update globals --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> commit ebbf206 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Wed Jul 9 15:52:32 2025 +0200 Update check-standard.yaml [skip ci] commit a65605f Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Wed Jul 9 09:13:24 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9011 [skip ci] commit 05b7e51 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Wed Jul 9 11:11:56 2025 +0200 Improve recist data (#75) * better arm repartition * improve date_inclusion calculation * improve documentation * example_rc: removed rctlsum_b & rctlmin * moved test-alexis.R to test-data.R * example_rc: add NTL yesno column `rcntlyn` NTL are NE mostly when non present * example_rc: fixed labels bind_rows removed them, added at last * improved documentation * example_rc: reformat * add .sample_yesno() * example_rc: improve workflow * example_rc: fix number of timepoints baseline was added * example_rc: removed unused variables * example_rc: minor reformat, renaming, and doc * example_rc: internalise things inside .simulate_one_patient() rctlsum_b + calculate rcdt from delai example_rc: simulate baseline inside .simulate_one_patient() also, decrease linearly (% of baseline) instead of exponentially (% of previous date) for simplicity * example_rc: revert baseline being not missing +response factor level management * example_rc: tweak default values * add tests on RECIST data (survival rates...) * add dev tests on RECIST (skipped, for viz only) * fix for check * add missing values * update doc * add NA/NE for real this time * update test for NE values --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> * remove use of `\(` shorthand * fix for check --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> commit a304e92 Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Fri Jul 11 16:58:20 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9014 [skip ci] commit 165566a Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Fri Jul 11 18:56:53 2025 +0200 new gr_officer_template() (#72) * new gr_officer_template() * doc * Update _pkgdown.yml * autoimport * fix for check * update globals --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> commit 63d0d15 Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Fri Jul 11 16:07:16 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9013 [skip ci] commit f95b324 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Fri Jul 11 18:06:07 2025 +0200 Implémenter les checks RECIST (#63) * WIP * recist_check: .add_to_recist_issues() instead of warn use an environment to centralize the issues instead of warning for each. + heavy reformat (should have made 2 commits...) * recist_check reorder * check_recist: .apply_recist_mapping() * check_lesion_number() * fixed .apply_recist_mapping * check_constancy * check_baseline_lesions * hotfix: add rtn objects * check_constancy: Response should be constant per date * check_baseline_lesions: baseline response should be missing * check_derived_columns * check_lesion_number() becomes check_target_lesions() * check_missing * check_target_response * fix check_missing * check_constancy: add nontarget_yn * check_target_response (for real) * check_global_response * .summarise_recist (for rc_short) * move utils to utils.R * check_global_response: fix arguments * .apply_recist_mapping: add mandatory variables * recist_issue: allow newline in message * backup commit * remove old code OLD_check_bare_recist OLD_check_response FIX_check_crf_structure * check_nontarget_response Patients with no Non-Target Lesions should not have a NTL response. * check_global_response: check NTL progressions Progressions due only to Non-Target Lesions are seldom and should be checked * remove legacy code * print.check_recist * remove unused grstat_env * gr_recist_mapping: remove unused variables * print.check_recist : nicer * document * check_recist: add mapping as attribute * rephrase most issue messages * recist_report() !!! * autoimport * check_derived_columns: conditional checks * improve documentation * Update _pkgdown.yml * recist_report: add title parameter * Delete test-recist.R * add globalVariables * fix for check * check_derived_columns: deduplicate checks * check_global_response: fix nlt-only progressions * template recist: improve options remove rownames, remove research * add assert_extension() * recist_report: split into HTML and XLSX * move to recist_utils.R * renamed to .recist_to_num() (and improved) * Remove rows posterior to the first progression * rewrote error messages * check that CR are terminal * check: Target Lesion should be <2 lymph nodes * signal dev state improves #69 * clean browser() reliquates * Update .gitignore * check_global_response: fix NA/NE * improve globalVariables() management * improve doc * fix for check * Update .gitignore * remove nonsensical check nadir can evolve over time, obviously * prefix internal check function with rc_ * add some tests * Update _pkgdown.yml * update doc --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> commit 48afe41 Author: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> Date: Fri Jul 11 14:05:35 2025 +0200 changed check action [skip ci] commit 33dd5ee Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Fri Jul 11 10:36:13 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9012 [skip ci] commit 68d87be Author: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> Date: Fri Jul 11 12:35:04 2025 +0200 Update check-standard.yaml commit 42db0f2 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Fri Jul 11 12:07:07 2025 +0200 add BOIN plot (#77) * add boin plot * autoimport * doc * add tests * Update _pkgdown.yml * fix for check * legend inside the gantt plot * clean unused variable * use boin$boundary_tab if full_boundary_tab is not available * improve documentation * update tests with newer example * fix for check * renamed data_patients columns + doc * test: check for unknown dose * typo * update tests * Update .gitignore * add experimental badge * update globals --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> commit ebbf206 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Wed Jul 9 15:52:32 2025 +0200 Update check-standard.yaml [skip ci] commit a65605f Author: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Date: Wed Jul 9 09:13:24 2025 +0000 Update dev version (Github Actions) Bump to v0.1.0.9011 [skip ci] commit 05b7e51 Author: Dan Chaltiel <dan.chaltiel@gmail.com> Date: Wed Jul 9 11:11:56 2025 +0200 Improve recist data (#75) * better arm repartition * improve date_inclusion calculation * improve documentation * example_rc: removed rctlsum_b & rctlmin * moved test-alexis.R to test-data.R * example_rc: add NTL yesno column `rcntlyn` NTL are NE mostly when non present * example_rc: fixed labels bind_rows removed them, added at last * improved documentation * example_rc: reformat * add .sample_yesno() * example_rc: improve workflow * example_rc: fix number of timepoints baseline was added * example_rc: removed unused variables * example_rc: minor reformat, renaming, and doc * example_rc: internalise things inside .simulate_one_patient() rctlsum_b + calculate rcdt from delai example_rc: simulate baseline inside .simulate_one_patient() also, decrease linearly (% of baseline) instead of exponentially (% of previous date) for simplicity * example_rc: revert baseline being not missing +response factor level management * example_rc: tweak default values * add tests on RECIST data (survival rates...) * add dev tests on RECIST (skipped, for viz only) * fix for check * add missing values * update doc * add NA/NE for real this time * update test for NE values --------- Co-authored-by: Dan Chaltiel <15105152+DanChaltiel@users.noreply.github.com> Revert "suite modif dan" This reverts commit 15a2cff.

Avant, utilisation de "Complete response" par exemple, mais dans le cas ou les données recist utilisait "CR" fonction ne marchait plus donc modif pour utiliser la valeur numérique qui est plus générale

DanChaltiel · 2025-08-29T10:57:28Z

Je viens de voir, le commit 946a74c n'est pas du tout RGPD parce qu'on cite le numéro du patient et le nom de l'étude.
Il faut donc qu'on fasse une nouvelle branche, avec une nouvelle PR et qu'on repart de là.
On va perdre l'historique des commits mais c'est pas grave du tout.
On fait ça ensemble à mon retour la semaine prochaine, ne t'inquiète pas!

Ajout de la vignette ORR

be1d976

AlexisCochard linked an issue Apr 29, 2025 that may be closed by this pull request

Vignette : Calcul des ORR #58

Open

4 tasks

DanChaltiel requested review from DanChaltiel and Lpierott April 30, 2025 08:34

Lpierott requested changes May 19, 2025

View reviewed changes

Lpierott reviewed May 19, 2025

View reviewed changes

Lpierott reviewed Jul 2, 2025

View reviewed changes

DanChaltiel mentioned this pull request Jul 8, 2025

Création d'un manuel pour l'utilisation de R en Recherche Clinique #60

Open

5 tasks

AlexisCochard added 14 commits July 23, 2025 15:19

modification texte suite commentaires livia

d146eae

maj du data.R par Dan

5bf6a5f

retrait de tout ce qui est lié au patient 129 de icarus

946a74c

Revert "maj du data.R par Dan"

48b99b2

This reverts commit 5bf6a5f.

suite modif dan

15a2cff

Fonction pour table ORR v0

c80f642

mise a jour de l'environnement avec la nouvelle fonction

92fc2d7

survfit_stack de dan

ff14792

ajout des cli_abort dans ORR_table + optimisation de la confirmation

fdd50ce

modif de dan

632fb01

maj des test et exemples

143d76e

Clean du code + ajout de cli_abort

32ce6ba

modif des cli_abort pour que ça marche

82ee36d

modif après test

001e307

DanChaltiel reviewed Jul 28, 2025

View reviewed changes

AlexisCochard added 6 commits July 29, 2025 15:18

Revert "modif de dan"

95c1633

This reverts commit 632fb01.

Revert "suite modif dan"

254b93d

This reverts commit 15a2cff.

documentation + test

58e8819

maj de calc_best_response avec ma methode

88a17a1

ajout de la confirmation ou non à calc_best_response

a01ec89

AlexisCochard added 10 commits August 4, 2025 11:04

documentation ORR_table

0550e9c

suppression code doublons de clac_best_reponse

88d5aac

optimisation dans ORR_table

e71b263

modif du as_flextable

2b1c24d

maj de la doc

696b9c9

ajout d'un distinct() pour supprimer les doublons

8377084

suppression de survfit_stack qui était apparu

02138da

modif pour prendre en compte les valeurs numériques

f0891bd

Avant, utilisation de "Complete response" par exemple, mais dans le cas ou les données recist utilisait "CR" fonction ne marchait plus donc modif pour utiliser la valeur numérique qui est plus générale

nettoyage

fb3055b

inversion des SD et PD dans la légende je sais pas pourquoi

7944ed8

		# Introduction

		This vignette illustrates the implementation of ORR analysis in R using `flextable` package. The ORR method is a standard approach in clinical research for estimating the percentage of people in a study or treatment group who have a partial response or complete response to the treatment within a certain period of time. A partial response is a decrease in the size of a tumor or in the amount of cancer in the body, and a complete response is the disappearance of all signs of cancer in the body. In a clinical trial, measuring the ORR is one way to see how well a new treatment works. Also called objective response rate.

		```

		Finally, we will use the `flextable` package for a table of results


		## Data Overview

		To calculate the ORR, we'll use the “recist” dataset.



		#Since we have 2 arms, we will first creae the part of the control

Ajout de la vignette ORR #67

Are you sure you want to change the base?

Ajout de la vignette ORR #67

Uh oh!

Conversation

AlexisCochard commented Apr 29, 2025

Uh oh!

DanChaltiel commented Apr 30, 2025

Uh oh!

Lpierott left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Lpierott Jul 1, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

DanChaltiel left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

DanChaltiel commented Aug 29, 2025

Uh oh!

Uh oh!

Lpierott Jul 1, 2025 •

edited

Loading