CTAS Validation
ctasval.RdThis function performs CTAS validation by generating anomalies, calculating scores, and summarizing performance metrics.
Usage
ctasval(
df,
fun_anomaly,
feats,
anomaly_degree = c(0, 0.5, 1, 5, 10, 50),
thresh = 1,
iter = 100,
n_sites = 3,
parallel = FALSE,
progress = TRUE,
default_minimum_timepoints_per_series = 3,
default_minimum_subjects_per_series = 3,
default_max_share_missing_timepoints_per_series = 0.5,
default_generate_change_from_baseline = FALSE,
autogenerate_timeseries = TRUE
)Arguments
- df
Data frame containing the study data.
- fun_anomaly
List of functions to apply to generate anomalies.
- feats
List of features to calculate for the timeseries.
- anomaly_degree
Vector of anomaly degrees to add. Default is c(0, 0.5, 1, 5, 10, 50).
- thresh
Threshold for classification. Default is 1.0.
- iter
Number of iterations to run. Default is 100.
- n_sites
Number of sites to generate. Default is 3.
- parallel
Logical indicating whether to run in parallel. Default is FALSE.
- progress
Logical indicating whether to show progress. Default is TRUE.
- default_minimum_timepoints_per_series
Minimum timepoints per series. Default is 3.
- default_minimum_subjects_per_series
Minimum subjects per series. Default is 3.
- default_max_share_missing_timepoints_per_series
Maximum share of missing timepoints per series. Default is 0.5.
- default_generate_change_from_baseline
Logical indicating whether to generate change from baseline. Default is FALSE.
- autogenerate_timeseries
Logical indicating whether to auto-generate timeseries. Default is TRUE.
Examples
df_prep <- prep_sdtm_lb(pharmaversesdtm::lb, pharmaversesdtm::dm, scramble = TRUE)
df_filt <- df_prep %>%
filter(parameter_id == "Alkaline Phosphatase")
ctas <- ctasval(
df = df_filt,
fun_anomaly = c(anomaly_average, anomaly_sd),
feats = c("average", "sd"),
parallel = FALSE,
iter = 1
)
ctas
#> $result
#> # A tibble: 12 × 9
#> anomaly_degree feats parameter_id TN FN FP TP tpr fpr
#> <dbl> <chr> <chr> <int> <int> <int> <int> <dbl> <dbl>
#> 1 0 average Alkaline Phosphat… 17 3 0 0 0 0
#> 2 0 sd Alkaline Phosphat… 17 3 0 0 0 0
#> 3 0.5 average Alkaline Phosphat… 17 3 0 0 0 0
#> 4 0.5 sd Alkaline Phosphat… 17 3 0 0 0 0
#> 5 1 average Alkaline Phosphat… 17 3 0 0 0 0
#> 6 1 sd Alkaline Phosphat… 17 3 0 0 0 0
#> 7 5 average Alkaline Phosphat… 17 3 0 0 0 0
#> 8 5 sd Alkaline Phosphat… 17 3 0 0 0 0
#> 9 10 average Alkaline Phosphat… 17 2 0 1 0.333 0
#> 10 10 sd Alkaline Phosphat… 17 3 0 0 0 0
#> 11 50 average Alkaline Phosphat… 17 3 0 0 0 0
#> 12 50 sd Alkaline Phosphat… 17 2 0 1 0.333 0
#>
#> $anomaly
#> # A tibble: 3,358 × 38
#> iter anomaly_degree fun_anomaly feats STUDYID DOMAIN subject_id LBSEQ
#> <int> <dbl> <list> <chr> <chr> <chr> <chr> <dbl>
#> 1 1 0 <fn> average CDISCPILOT01 LB sample_si… 2
#> 2 1 0 <fn> average CDISCPILOT01 LB sample_si… 39
#> 3 1 0 <fn> average CDISCPILOT01 LB sample_si… 74
#> 4 1 0 <fn> average CDISCPILOT01 LB sample_si… 104
#> 5 1 0 <fn> average CDISCPILOT01 LB sample_si… 134
#> 6 1 0 <fn> average CDISCPILOT01 LB sample_si… 164
#> 7 1 0 <fn> average CDISCPILOT01 LB sample_si… 199
#> 8 1 0 <fn> average CDISCPILOT01 LB sample_si… 229
#> 9 1 0 <fn> average CDISCPILOT01 LB sample_si… 259
#> 10 1 0 <fn> average CDISCPILOT01 LB sample_si… 2
#> # ℹ 3,348 more rows
#> # ℹ 30 more variables: LBTESTCD <chr>, LBTEST <chr>, LBCAT <chr>,
#> # LBORRES <chr>, LBORRESU <chr>, LBORNRLO <chr>, LBORNRHI <chr>,
#> # LBSTRESC <chr>, LBSTRESN <dbl>, LBSTRESU <chr>, LBSTNRLO <dbl>,
#> # LBSTNRHI <dbl>, LBNRIND <chr>, LBBLFL <chr>, VISITNUM <dbl>, VISIT <chr>,
#> # VISITDY <dbl>, LBDTC <chr>, LBDY <dbl>, timepoint_rank <dbl>,
#> # timepoint_1_name <chr>, result <dbl>, parameter_id <chr>, …
#>
#> attr(,"class")
#> [1] "ctasval_aggregated"