Confounder-Adjusted Survival Curves and Cumulative Incidence Functions

Estimate and plot confounder-adjusted survival curves using either 'Direct Adjustment', 'Direct Adjustment with Pseudo-Values', various forms of 'Inverse Probability of Treatment Weighting', two forms of 'Augmented Inverse Probability of Treatment Weighting', 'Empirical Likelihood Estimation' or 'Targeted Maximum Likelihood Estimation'. Also includes a significance test for the difference between two adjusted survival curves and the calculation of adjusted restricted mean survival times. Additionally enables the user to estimate and plot cause-specific confounder-adjusted cumulative incidence functions in the competing risks setting using the same methods (with some exceptions). For details, see Denz et. al (2023) .

Calculate the Care Density or Fragmented Care Density Given a Patient-Sharing Network

Given a patient-sharing network, calculate either the classic care density as proposed by Pollack et al. (2013) or the fragmented care density as proposed by Engels et al. (2024) . By utilizing the 'igraph' and 'data.table' packages, the provided functions scale well for very large graphs.

Open GenBank Files

Opens complete record(s) with .gb extension from the NCBI/GenBank Nucleotide database and returns a list containing shaped record(s). These kind of files contains detailed records of DNA samples (locus, organism, type of sequence, source of the sequence...). An example of record can be found at < https://www.ncbi.nlm.nih.gov/nuccore/HE799070>.

Distances on Directed Graphs

Distances on dual-weighted directed graphs using priority-queue shortest paths (Padgham (2019) ). Weighted directed graphs have weights from A to B which may differ from those from B to A. Dual-weighted directed graphs have two sets of such weights. A canonical example is a street network to be used for routing in which routes are calculated by weighting distances according to the type of way and mode of transport, yet lengths of routes must be calculated from direct distances.

Imputation of High-Dimensional Count Data using Side Information

Analysis, imputation, and multiple imputation of count data using covariates. LORI uses a log-linear Poisson model where main row and column effects, as well as effects of known covariates and interaction terms can be fitted. The estimation procedure is based on the convex optimization of the Poisson loss penalized by a Lasso type penalty and a nuclear norm. LORI returns estimates of main effects, covariate effects and interactions, as well as an imputed count table. The package also contains a multiple imputation procedure. The methods are described in Robin, Josse, Moulines and Sardy (2019) .

Variable Selection Using Random Forests

Three steps variable selection procedure based on random forests. Initially developed to handle high dimensional data (for which number of variables largely exceeds number of observations), the package is very versatile and can treat most dimensions of data, for regression and supervised classification problems. First step is dedicated to eliminate irrelevant variables from the dataset. Second step aims to select all variables related to the response for interpretation purpose. Third step refines the selection by eliminating redundancy in the set of variables selected by the second step, for prediction purpose. Genuer, R. Poggi, J.-M. and Tuleau-Malot, C. (2015) < https://journal.r-project.org/articles/RJ-2015-018/>.

Fast Symbolic Multivariate Polynomials

Fast manipulation of symbolic multivariate polynomials using the 'Map' class of the Standard Template Library. The package uses print and coercion methods from the 'mpoly' package but offers speed improvements. It is comparable in speed to the 'spray' package for sparse arrays, but retains the symbolic benefits of 'mpoly'. To cite the package in publications, use Hankin 2022 . Uses 'disordR' discipline.

The Hyperdirichlet Distribution, Mark 2

A suite of routines for the hyperdirichlet distribution and reified Bradley-Terry; supersedes the 'hyperdirichlet' package; uses 'disordR' discipline . To cite in publications please use Hankin 2017 , and for Generalized Plackett-Luce likelihoods use Hankin 2024 .

Interface Between 'GRASS' Geographical Information System and 'R'

An interface between the 'GRASS' geographical information system ('GIS') and 'R', based on starting 'R' from within the 'GRASS' 'GIS' environment, or running a free-standing 'R' session in a temporary 'GRASS' location; the package provides facilities for using all 'GRASS' commands from the 'R' command line. The original interface package for 'GRASS 5' (2000-2010) is described in Bivand (2000) and Bivand (2001) < https://www.r-project.org/conferences/DSC-2001/Proceedings/Bivand.pdf>. This was succeeded by 'spgrass6' for 'GRASS 6' (2006-2016) and 'rgrass7' for 'GRASS 7' (2015-present). The 'rgrass' package modernizes the interface for 'GRASS 8' while still permitting the use of 'GRASS 7'.

Extension to 'tmap' for Creating Network Visualizations

Provides functions for visualizing networks with 'tmap'. It supports 'sfnetworks' objects natively but is not limited to them. Useful for adding network layers such as edges and nodes to 'tmap' maps. More features may be added in future versions.