calc_bal_stats change

omitting weighting on calc_bal_stats histories for which exposure is all 0s (in which case col_w_smd() cannot calculate bal stats)

Author: istallworthy

R/assessBalance.R

@@ -460,7 +460,7 @@ summary.devMSM_bal_stats <- function(object, i = NA, save.out = FALSE, ...) {
         if (!is.na(i)) {
           sprintf("Imbalanced Covariates for imputation %s using `%s`", i, weight_method)
         } else {
-          sprintf("Imbalanced Covariates Averaging Across Imputed Datasetrs using `%s`", weight_method)
+          sprintf("Imbalanced Covariates Averaging Across Imputed Datasets using `%s`", weight_method)
         }
       } else {
         if (!is.na(i)) {

R/calc_bal_stats.R

@@ -122,7 +122,16 @@ calc_bal_stats <- function(data, obj, weights = NULL, balance_thresh = NULL, imp
           prop_sum$history[prop_sum$Freq == 1 | prop_sum$Freq == 0]
         )
       } # ends hist exc
-
+      
+      # added by IS 4/8/26 to skip histories for which exposures are all 0s (cannot calc bal stats)
+      omitted_histories <- Filter(
+        function(h) {
+          which_idx <- (history == h)
+          sum(data[[exposure_name]][which_idx]) == 0
+        },
+        setdiff(prop_sum$history, omitted_histories)
+      )
+      
       # finding balance by history
       # if weighted, use IPTW weights from weightitmsm and weight by history
       history_bal_stats <- sapply(
@@ -137,10 +146,15 @@ calc_bal_stats <- function(data, obj, weights = NULL, balance_thresh = NULL, imp
           )
         }
       )
-
+      
+      # IS amended to exclude ommitted histories
       # getting weighted mean across histories (cols), weighting by proportion of those w/ that same history
-      weighted_bal_stats <- history_bal_stats %*% prop_sum$prop
-
+      # weighted_bal_stats <- history_bal_stats %*% prop_sum$prop
+      keep <- !(prop_sum$history %in% omitted_histories)
+      props <- as.numeric(prop_sum$prop[keep])
+      weighted_bal_stats <- history_bal_stats %*% props
+      
+      
       if (exposure_type == "continuous") {
         # unweighted covar sd *
         s1 <- cobalt::col_w_sd(bal_vars)

vignettes/Data_Requirements.qmd

@@ -424,7 +424,7 @@ The user can specify an imputation method compatible with `mice()` (see above).
 
 The user may also specify any additional inputs accepted by `mice::mice()` and we advise consulting the <a href:="https://www.rdocumentation.org/packages/mice/versions/3.16.0/topics/mice">\[*mice* documentation\]</a> for more information.
 
-The user can also indicate if they have already created imputed datasets from this function and wish to read them in (`read_imps_from_file = TRUE` rather than recreate them (default).
+The user can also indicate if they have already created imputed datasets from this function and wish to read them in (`read_imps_from_file = TRUE`) rather than recreate them (default).
 
 For this example, we create 2 imputed datasets using the default random forest method and 0 iterations (just for illustrative purposes), set a seed for reproducibility, and assign the output to `data` for use with *devMSMs*. This code takes some time to run. (Note: given the challenges of imputing data from .rda files, we have set `m` = 2 and `maxit` = 0 here just for illustrative purposes. We recommend setting both `m` = 5 and `maxit` = 5 (*mice* default) when running data.)