mscpredmodel: mscpredmodel: Multiple Score Comparison of Prediction Models

In srhaile/mscpredmodel: Multiple Score Comparison of Prediction Models with Network Meta-Analysis

Description

The mscpredmodel package provides a number of functions to facilitate multiple score comparison (MSC), a network meta-analytic approach to comparing multiple prediction models or prognostic scores using individual patient data (IPD).

Main Functions

The main mscpredmodel functions are as follows, in approximately the order they might be used in a data analysis:

msc_sample_data

Produces a simulated dataset in order to try out the package

get_bs_samples

Generate bootstrap samples, stratified by cohort

compute_performance

Compute a performance measure for each of the scores, stratified by cohort, in each of the bootstrap samples. One such performance measure would be, for example, calibration_slope

aggregate_performance

Produces the aggregated performance and its empirically estimated variance-covariance matrix for each cohort, to be used in the (in)consistency models.

consistency

Computes the (in)consistency model.

msc_full

Based on the raw performance estimates from compute_performance, compute direct (msc_direct), indirect (msc_indirect), or full network (msc_network) pairwise comparisons. Using msc_full, all three sets of pairwise estimates are computed at once.

srhaile/mscpredmodel documentation built on Sept. 13, 2019, 3:44 p.m.