This task extends mlr3::Task and mlr3::TaskSupervised for competing risks survival analysis. The target consists of a survival time and an event indicator. Event codes must be non-negative integers in \((0, 1, 2, ..., K)\). \(0\) denotes censoring, and positive integers denote distinct event causes. Each row represents one observation.
Predefined tasks are stored in mlr3::mlr_tasks.
The task_type is set to "cmprsk".
Details
The following design choices apply to this task:
Only right-censoring is currently supported.
Tasks must contain at least two event causes, i.e., \(K \geq 2\), encoded as consecutive integers \(1, 2, ..., K\).
It is advised to use stratified resampling to reduce the risk of creating training splits with fewer than two causes, which may cause issues during model training (and later prediction). Task filtering specifically issues a warning when the number of competing events is reduced but remains >= 2, and an error if fewer than two causes remain.
See also
Other Task:
mlr_tasks_pbc
Super classes
mlr3::Task -> mlr3::TaskSupervised -> TaskCompRisks
Active bindings
cens_type(
character(1))
Returns the type of censoring.Currently, only
"right"censoring type is supported. The API might change in the future to support left and interval censoring.cmp_events(
character())
Returns the competing event names:"1","2", ...,"K", in that order.
Methods
Inherited methods
mlr3::Task$add_strata()mlr3::Task$cbind()mlr3::Task$data()mlr3::Task$droplevels()mlr3::Task$format()mlr3::Task$head()mlr3::Task$help()mlr3::Task$levels()mlr3::Task$materialize_view()mlr3::Task$missings()mlr3::Task$print()mlr3::Task$rbind()mlr3::Task$rename()mlr3::Task$select()mlr3::Task$set_col_roles()mlr3::Task$set_levels()mlr3::Task$set_row_roles()
TaskCompRisks$new()
Creates a new instance of this R6 class.
Usage
TaskCompRisks$new(
id,
backend,
time = "time",
event = "event",
label = NA_character_
)Arguments
id(
character(1))
Identifier for the new instance.backend(mlr3::DataBackend)
Either a DataBackend, or any object which is convertible to a DataBackend withas_data_backend(). E.g., adata.frame()will be converted to a DataBackendDataTable.time(
character(1))
Name of the column for outcome time.event(
character(1))
Name of column that holds the event indicator. \(0\) corresponds to censoring, values \(> 0\) correspond to different competing events (or causes).label(
character(1))
Label for the new instance.
TaskCompRisks$truth()
True response for specified row_ids. This is the multi-state format
using Surv with the event target column as a factor.
Defaults to all rows with role "use".
Arguments
rows(
integer())
Row indices.
TaskCompRisks$formula()
Creates a formula for competing risk models with survival::Surv() on
the LHS (left hand side).
TaskCompRisks$times()
Returns the (unsorted) outcome times.
Arguments
rows(
integer())
Row indices.
TaskCompRisks$event()
Returns the event indicators. \(0\) denotes censoring, and positive integers denote distinct event causes.
Arguments
rows(
integer())
Row indices.
TaskCompRisks$unique_times()
Returns the sorted unique outcome times.
Arguments
rows(
integer())
Row indices.
TaskCompRisks$unique_event_times()
Returns the sorted unique event outcome times (by any cause).
Arguments
rows(
integer())
Row indices.
TaskCompRisks$aalen_johansen()
Calls survival::survfit() to calculate the Aalen–Johansen estimator.
Arguments
strata(
character())
Stratification variables to use.rows(
integer())
Subset of row indices....(any)
Additional arguments passed down tosurvival::survfit.formula().
TaskCompRisks$cens_prop()
Returns the proportion of censored observations for this competing risks task.
Arguments
rows(
integer())
Row indices.
TaskCompRisks$filter()
Subsets the task, keeping only the rows specified via the row ids rows.
This operation mutates the task in-place.
A warning is thrown if the filtering results in fewer competing events
than the original task.
An error is thrown if fewer than two competing events remain after filtering.
Arguments
rows(
integer())
Row indices.
Examples
library(mlr3)
task = tsk("pbc")
# Time and event target columns
task$target_names
#> [1] "time" "status"
# Feature names
task$feature_names
#> [1] "age" "albumin" "alk.phos" "ascites" "ast" "bili"
#> [7] "chol" "copper" "edema" "hepato" "platelet" "protime"
#> [13] "sex" "spiders" "stage" "trig" "trt"
# Censoring type
task$cens_type
#> [1] "right"
# survival data
task$formula(c("age", "sex")) # formula with survival::Surv() on LHS
#> Surv(time, factor(status, levels = c(0, 1, 2))) ~ age + sex
#> <environment: namespace:survival>
task$truth() # survival::Surv() object
#> [1] 13:2 147+ 33:2 63:2 49:1 60+ 81:2 78:2 1:2 123:2 9:2 117+
#> [13] 117:2 120+ 25:2 4:2 139+ 44:2 113+ 22:2 8:2 134:2 135+ 47:2
#> [25] 2:2 18:2 148+ 10:2 126:2 148+ 104:2 129+ 93:2 118+ 7:2 106:2
#> [37] 75:2 149+ 112:2 74:2 84+ 145+ 85:2 126:2 78:2 47:2 44:2 60:2
#> [49] 107:2 73:2 143+ 139+ 101:2 28:2 48:2 131+ 137:2 90:2 132+ 38:2
#> [61] 137+ 137+ 137+ 60:2 39:2 2:2 10:2 55:2 121+ 29:2 83:2 117:2
#> [73] 133+ 132+ 110:2 54:2 6:2 80+ 57:2 88:2 15:2 12:2 128+ 24:2
#> [85] 20:2 125+ 125+ 18:2 117+ 101+ 3:2 101:2 101:1 111+ 84:2 82+
#> [97] 69:2 77:1 113:2 32:2 111:2 112+ 109+ 35:2 75:2 16:2 66:1 6:2
#> [109] 108+ 100+ 81:1 106+ 46:2 27:2 96+ 91:2 103+ 103+ 101+ 98+
#> [121] 42:2 69+ 100+ 100+ 79:2 25:2 30:2 97+ 98+ 46:2 25:2 94+
#> [133] 37:2 94+ 4:2 87+ 28:2 93+ 81:1 50:2 91+ 91+ 6:2 67:2
#> [145] 35:2 89+ 55:2 39:2 88+ 75+ 87+ 86+ 85+ 84+ 84+ 83+
#> [157] 73:1 31:2 83+ 51:2 24:2 76+ 80+ 7:2 80+ 26:2 80+ 76+
#> [169] 77+ 77+ 77+ 52+ 76+ 75+ 74+ 72+ 68:2 74+ 29:2 72+
#> [181] 73+ 71+ 71+ 58:2 35:2 25:2 70+ 40:2 67+ 19:2 10:2 63+
#> [193] 66+ 64+ 32:2 64+ 11:2 65+ 38:2 64+ 63+ 58+ 61+ 62+
#> [205] 61+ 22:2 60+ 27:1 59+ 30:2 55:2 58+ 47:1 24:1 58+ 58+
#> [217] 58+ 47+ 58+ 57+ 24:1 56+ 55+ 53+ 55+ 53+ 54+ 54+
#> [229] 42:1 50:1 35:1 53+ 5:2 39:2 44+ 51+ 51+ 50+ 51+ 44+
#> [241] 48+ 47+ 46+ 47+ 46+ 1:2 47+ 33+ 46+ 46+ 46+ 40+
#> [253] 35:1 26:2 44+ 29:1 43+ 43+ 42+ 28:1 43+ 17:1 42+ 42+
#> [265] 42+ 41+ 41+ 40+ 39+ 39+ 37+ 37+ 32+ 30+ 27+ 25+
task$times() # (unsorted) times
#> [1] 13 147 33 63 49 60 81 78 1 123 9 117 117 120 25 4 139 44
#> [19] 113 22 8 134 135 47 2 18 148 10 126 148 104 129 93 118 7 106
#> [37] 75 149 112 74 84 145 85 126 78 47 44 60 107 73 143 139 101 28
#> [55] 48 131 137 90 132 38 137 137 137 60 39 2 10 55 121 29 83 117
#> [73] 133 132 110 54 6 80 57 88 15 12 128 24 20 125 125 18 117 101
#> [91] 3 101 101 111 84 82 69 77 113 32 111 112 109 35 75 16 66 6
#> [109] 108 100 81 106 46 27 96 91 103 103 101 98 42 69 100 100 79 25
#> [127] 30 97 98 46 25 94 37 94 4 87 28 93 81 50 91 91 6 67
#> [145] 35 89 55 39 88 75 87 86 85 84 84 83 73 31 83 51 24 76
#> [163] 80 7 80 26 80 76 77 77 77 52 76 75 74 72 68 74 29 72
#> [181] 73 71 71 58 35 25 70 40 67 19 10 63 66 64 32 64 11 65
#> [199] 38 64 63 58 61 62 61 22 60 27 59 30 55 58 47 24 58 58
#> [217] 58 47 58 57 24 56 55 53 55 53 54 54 42 50 35 53 5 39
#> [235] 44 51 51 50 51 44 48 47 46 47 46 1 47 33 46 46 46 40
#> [253] 35 26 44 29 43 43 42 28 43 17 42 42 42 41 41 40 39 39
#> [271] 37 37 32 30 27 25
task$event() # event indicators (0 = censored, >0 = different causes)
#> [1] 2 0 2 2 1 0 2 2 2 2 2 0 2 0 2 2 0 2 0 2 2 2 0 2 2 2 0 2 2 0 2 0 2 0 2 2 2
#> [38] 0 2 2 0 0 2 2 2 2 2 2 2 2 0 0 2 2 2 0 2 2 0 2 0 0 0 2 2 2 2 2 0 2 2 2 0 0
#> [75] 2 2 2 0 2 2 2 2 0 2 2 0 0 2 0 0 2 2 1 0 2 0 2 1 2 2 2 0 0 2 2 2 1 2 0 0 1
#> [112] 0 2 2 0 2 0 0 0 0 2 0 0 0 2 2 2 0 0 2 2 0 2 0 2 0 2 0 1 2 0 0 2 2 2 0 2 2
#> [149] 0 0 0 0 0 0 0 0 1 2 0 2 2 0 0 2 0 2 0 0 0 0 0 0 0 0 0 0 2 0 2 0 0 0 0 2 2
#> [186] 2 0 2 0 2 2 0 0 0 2 0 2 0 2 0 0 0 0 0 0 2 0 1 0 2 2 0 1 1 0 0 0 0 0 0 1 0
#> [223] 0 0 0 0 0 0 1 1 1 0 2 2 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 1 2 0 1 0 0 0
#> [260] 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0
task$unique_times() # sorted unique times
#> [1] 1 2 3 4 5 6 7 8 9 10 11 12 13 15 16 17 18 19
#> [19] 20 22 24 25 26 27 28 29 30 31 32 33 35 37 38 39 40 41
#> [37] 42 43 44 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
#> [55] 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78
#> [73] 79 80 81 82 83 84 85 86 87 88 89 90 91 93 94 96 97 98
#> [91] 100 101 103 104 106 107 108 109 110 111 112 113 117 118 120 121 123 125
#> [109] 126 128 129 131 132 133 134 135 137 139 143 145 147 148 149
task$unique_event_times() # sorted unique event times (from any cause)
#> [1] 1 2 3 4 5 6 7 8 9 10 11 12 13 15 16 17 18 19 20
#> [20] 22 24 25 26 27 28 29 30 31 32 33 35 37 38 39 40 42 44 46
#> [39] 47 48 49 50 51 54 55 57 58 60 63 66 67 68 69 73 74 75 77
#> [58] 78 79 81 83 84 85 88 90 91 93 101 104 106 107 110 111 112 113 117
#> [77] 123 126 134 137
task$cens_prop() # proportion of censored observations
#> [1] 0.5326087
# Aalen-Johansen estimator
task$aalen_johansen(strata = "sex")
#> Call: survfit(formula = f, data = data)
#>
#> n nevent rmean se(rmean)*
#> sex=m, (s0) 34 0 73.485839 8.777244
#> sex=f, (s0) 242 0 93.714998 3.738137
#> sex=m, 1 34 3 10.556980 5.624320
#> sex=f, 1 242 15 7.654985 1.904137
#> sex=m, 2 34 21 64.957181 9.152039
#> sex=f, 2 242 90 47.630017 3.733828
#> *restricted mean time in state (max time = 149 )
# Causes
task$cmp_events
#> [1] "1" "2"