LMScore: Methods (AUCt, Brier) to score the predictive performance of...

In anyafries/dynLM: Dynamic w-year risk predictions from landmark time points

Methods (AUCt, Brier) to score the predictive performance of dynamic risk markers from LM super models

Description

Methods (AUCt, Brier) to score the predictive performance of dynamic risk markers from LM super models

Usage

LMScore(
  preds,
  formula,
  metrics = c("auc", "brier"),
  cause,
  tLM,
  unit,
  split.method,
  B,
  ...
)

Arguments

preds	A named list of prediction models, where allowed entries are outputs from predLMrisk
formula	A survival or event history formula. The left hand side is used to compute the expected event status. If none is given, it is obtained from the prediction object.
metrics	Character vector specifying which metrics to apply. Choices are "auc" and "brier". Case matters.
cause	Cause of interest if considering competing risks
tLM	Landmark times for which scores must be given. These must be a subset of LM times used during the prediction
unit	Time unit for window of prediction, e.g., "year", "month", etc. Used for printing results.
split.method	Method for cross-validation. Right now, as in riskRegression, the only option is bootcv.
B	The number of bootstap steps for cross-validation.
...	Additional arguments to pass to Score (riskRegression package)

Details

See the Github for example code

Value

An object of class "LMScore", which has components:

• auct: dataframe containing time-dependent auc information if "auc" was a metric

• briert: dataframe containing time-dependent brier score if "brier" was a metric

anyafries/dynLM documentation built on July 26, 2022, 12:17 a.m.