README.md

In SkadiEye/deepTL: Deep Treatment Learning

Deep Treatment Learning (deepTL)

Deep Treatment Learning (deepTL) is an R packages written in S4 class, designed for,

• PermFIT: Permutation-based Feature Importance Test, a permutation-based feature importance test scheme for black-box models (DNNs, support vector machines, random forests, etc) [Manuscript submitted] [example]
• deepTL: Deep Treatment Learning, an efficient semiparametric framework coupled with ensemble DNNs for adjusting complex confounding [Manuscript submitted] [example]
• EndLot: ENsemble Decision Learning Optimal Treatment, a DNN-based method for optimal individualized treatment learning (Paper: Mi et al. (2019)) [example]

You may also use it for,

• DNN: Easy implementation for feed-forward fully-connected deep neural networks
• Bagging: Bootstrap aggregating for DNN models, with an automatic scheme to select the optimal subset of DNNs (details in paper: Mi et al. (2019))
• [example]

Installation

• System requirement: Rtools (Windows); None (MacOS/Linux)

• In R:

devtools::install_github("SkadiEye/deepTL")

References

Mi, X., Zou, F., and Zhu, R. (2019), “Bagging and deep learning in optimal individualized treatment rules,” *Biometrics*, Wiley Online Library, 75, 674–684.

SkadiEye/deepTL documentation built on Nov. 17, 2022, 1:41 p.m.