All vignettes can be found as articles on the website https://alexpate30.github.io/calibmsm/index.html.
The Overview vignette is a guide on how to use the package by running through a clinical example. This should be used as the starting point for understanding how to use calibmsm. This vignette will be submitted for publication when it is ready and a reference will be added here.
The BLR-IPCW-manual-bootstrap vignette showcases how to estimate a confidence interval manually using bootstrapping, and how to manually specify a function for estimating the inverse probability of censoring weights.
The Calibration-curves-estimated-with-loess-smoothers vignette repeats the analyses from the Overview vignette using loess smoothers to estimate the calibration curves instead of restricted cubic splines.
The Sensitivity-analysis-for-IPCWs vignette explores whether the assumptions of the BLR-IPCW method are met in the illustrative example in the Overview vignette. In particular, the sensitivity of the estimated calibration curves to the type of model (cox proportional hazards or flexible parametric) used to estimate the weights is considered, and how bias may occur when the censoring mechanism changes depending on outcome state occupancy.
The Comparison-in-competing-risks-setting vignette uses calibmsm to assess the calibration of a standard competing risk model and compares results with the more commonly used graphical calibration curves.