# SurvBregDiv **Transfer learning for time-to-event modelling via Bregman divergence.** `SurvBregDiv` enables principled borrowing of external information when fitting Cox proportional hazards or nested case–control (NCC) models, through a unified Bregman-divergence framework that accommodates population heterogeneity between internal and external cohorts. > #### Using SurvBregDiv with an AI assistant? > > An AI-optimized reference is published at > **** (following the > [llms.txt](https://llmstxt.org/) convention). Point your AI at that > URL, or paste its contents into the chat, to give the assistant a > compact map of the package — decision tree, parameter reference, > worked examples, and common pitfalls — without ingesting the full > website. ## Installation ``` r # CRAN install.packages("SurvBregDiv") # Development version from GitHub remotes::install_github("UM-KevinHe/SurvBregDiv") ``` Requires R ≥ 4.0. ## Documentation - **Tutorials and methodology**: - **Function reference**: ## Getting help The package is under active development; please report issues or unexpected behavior to any of the maintainers: - Yubo Shao — - Junyi Qiu — - Kevin He — # Package index ## Low-Dimensional Individual-Level Internal–External Integration Models that integrate individual-level external data via composite likelihood weighting, for both full-cohort Cox and nested case-control (NCC) designs. - [`cox_indi()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cox_indi.md) : Cox Proportional Hazards Model Integrated with External Individual-level Information - [`cv.cox_indi()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.cox_indi.md) : Cross-Validated cox_indi to Tune etas - [`ncc_indi()`](https://um-kevinhe.github.io/SurvBregDiv/reference/ncc_indi.md) : Conditional Logistic Regression with Individual-level External Data (CLR-Indi) - [`cv.ncc_indi()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.ncc_indi.md) : Cross-Validated CLR with Individual-Level External Data ## Low-Dimensional KL Divergence-Based Integration Models that integrate external coefficient summaries via Kullback–Leibler divergence penalization, for both full-cohort Cox and nested case-control (NCC) designs. - [`coxkl()`](https://um-kevinhe.github.io/SurvBregDiv/reference/coxkl.md) : Cox Proportional Hazards Model with KL Divergence for Data Integration - [`cv.coxkl()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.coxkl.md) : Cross-Validated Cox–KL to Tune the Integration Parameter (eta) - [`bopt.coxkl()`](https://um-kevinhe.github.io/SurvBregDiv/reference/bopt.coxkl.md) : Bayesian Optimization for the Cox–KL Integration Parameter (eta) - [`coxkl_ties()`](https://um-kevinhe.github.io/SurvBregDiv/reference/coxkl_ties.md) : Cox Proportional Hazards Model with KL Divergence for Data Integration (Ties Handling) - [`cv.coxkl_ties()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.coxkl_ties.md) : Cross-Validated Cox–KL with Ties Handling to Tune the Integration Parameter (eta) - [`ncckl()`](https://um-kevinhe.github.io/SurvBregDiv/reference/ncckl.md) : Conditional Logistic Regression with KL Divergence (CLR-KL) - [`cv.ncckl()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.ncckl.md) : Cross-Validated Conditional Logistic Regression with KL Integration ## Low-Dimensional Mahalanobis Distance-Based Integration Models that integrate external coefficient and curvature summaries via Mahalanobis distance penalization, for both full-cohort Cox and matched case-control (NCC) designs. - [`cox_MDTL()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cox_MDTL.md) : Cox Proportional Hazards Model with Mahalanobis Distance Transfer Learning - [`cv.cox_MDTL()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.cox_MDTL.md) : Cross-Validation for Cox MDTL Model - [`ncc_MDTL()`](https://um-kevinhe.github.io/SurvBregDiv/reference/ncc_MDTL.md) : Conditional Logistic Regression with Mahalanobis Distance Transfer Learning (CLR-MDTL) - [`cv.ncc_MDTL()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.ncc_MDTL.md) : Cross-Validated CLR with Mahalanobis Distance Transfer Learning ## High-Dimensional Individual-Level Internal–External Integration Penalized models with Elastic Net regularization that integrate individual-level external data, for both full-cohort Cox and nested case-control (NCC) designs. - [`cox_indi_enet()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cox_indi_enet.md) : Cox Proportional Hazards Model Integrated with External Individual-level Data and Elastic Net Penalty - [`cv.cox_indi_enet()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.cox_indi_enet.md) : Cross-Validation for Cox Model Integrated with External Individual-level Data and Elastic Net Penalty - [`ncc_indi_enet()`](https://um-kevinhe.github.io/SurvBregDiv/reference/ncc_indi_enet.md) : Conditional Logistic Regression with Individual-level External Data and Elastic Net Penalty (CLR-Indi-ENet) - [`cv.ncc_indi_enet()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.ncc_indi_enet.md) : Cross-Validated CLR with Individual-Level External Data and Elastic Net Penalty ## High-Dimensional KL Divergence-Based Integration Penalized models with Ridge, Lasso, or Elastic Net regularization that integrate external coefficient summaries via KL divergence, for both full-cohort Cox and nested case-control (NCC) designs. - [`coxkl_ridge()`](https://um-kevinhe.github.io/SurvBregDiv/reference/coxkl_ridge.md) : Cox Proportional Hazards Model with Ridge Penalty and External Information - [`coxkl_enet()`](https://um-kevinhe.github.io/SurvBregDiv/reference/coxkl_enet.md) : Cox Proportional Hazards Model with KL Divergence and Elastic Net Penalty - [`cv.coxkl_ridge()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.coxkl_ridge.md) : Cross-Validation for CoxKL Ridge Model (Tuning Eta and Lambda) - [`cv.coxkl_enet()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.coxkl_enet.md) : Cross-Validation for CoxKL Model with Elastic Net & Lasso Penalty - [`ncckl_enet()`](https://um-kevinhe.github.io/SurvBregDiv/reference/ncckl_enet.md) : Conditional Logistic Regression with KL Divergence and Elastic Net Penalty (CLR-KL-ENet) - [`cv.ncckl_enet()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.ncckl_enet.md) : Cross-Validated CLR-KL with Elastic Net Penalty ## High-Dimensional Mahalanobis Distance-Based Integration Penalized models with Ridge, Lasso, or Elastic Net regularization that integrate external coefficient and curvature summaries via Mahalanobis distance, for both full-cohort Cox and nested case-control (NCC) designs. - [`cox_MDTL_ridge()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cox_MDTL_ridge.md) : Cox MDTL with Ridge Regularization - [`cox_MDTL_enet()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cox_MDTL_enet.md) : Fit Cox Model with Multi-Domain Transfer Learning and Elastic Net Penalty - [`cv.cox_MDTL_ridge()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.cox_MDTL_ridge.md) : Cross-Validation for Cox MDTL with Ridge Regularization - [`cv.cox_MDTL_enet()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.cox_MDTL_enet.md) : Cross-Validation for Cox MDTL with Elastic Net Regularization - [`ncc_MDTL_enet()`](https://um-kevinhe.github.io/SurvBregDiv/reference/ncc_MDTL_enet.md) : Conditional Logistic Regression with Mahalanobis Distance Transfer Learning and Elastic Net (CLR-MDTL-ENet) - [`cv.ncc_MDTL_enet()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.ncc_MDTL_enet.md) : Cross-Validated CLR with Mahalanobis Distance Transfer Learning and Elastic Net Penalty ## Variable Selection and Stability for High-Dimensional Models - [`variable_importance()`](https://um-kevinhe.github.io/SurvBregDiv/reference/variable_importance.md) : Bootstrap Variable Importance via Selection Frequency - [`coxkl_enet.StabSelect()`](https://um-kevinhe.github.io/SurvBregDiv/reference/coxkl_enet.StabSelect.md) : Stability Selection for KL-Integrated Cox Elastic-Net Models - [`cox_MDTL_enet.StabSelect()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cox_MDTL_enet.StabSelect.md) : Stability Selection for MDTL-Integrated Cox Elastic-Net Models ## Bagging for High-Dimensional Models - [`coxkl_enet_bagging()`](https://um-kevinhe.github.io/SurvBregDiv/reference/coxkl_enet_bagging.md) : Bagging for KL-Integrated Cox Elastic-Net Models - [`cox_MDTL_enet_bagging()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cox_MDTL_enet_bagging.md) : Bagging for MDTL-Integrated Cox Elastic-Net Models ## Multi-Source Integration - [`coxkl_enet.multi()`](https://um-kevinhe.github.io/SurvBregDiv/reference/coxkl_enet.multi.md) : Multi-Source Integration for KL-Integrated Cox Elastic-Net Models ## Plotting Functions - [`cv.plot()`](https://um-kevinhe.github.io/SurvBregDiv/reference/cv.plot.md) : Plot Cross-Validation Results vs Eta - [`plot(`*``*`)`](https://um-kevinhe.github.io/SurvBregDiv/reference/plot.coxkl.md) : Plot Validation Results for coxkl Object - [`plot(`*``*`)`](https://um-kevinhe.github.io/SurvBregDiv/reference/plot.coxkl_ridge.md) : Plot Validation Results for coxkl_ridge Object - [`plot(`*``*`)`](https://um-kevinhe.github.io/SurvBregDiv/reference/plot.coxkl_enet.md) : Plot Validation Results for coxkl_enet Object - [`plot(`*``*`)`](https://um-kevinhe.github.io/SurvBregDiv/reference/plot.cox_MDTL.md) : Plot Validation Results for Cox_MDTL Object - [`plot(`*``*`)`](https://um-kevinhe.github.io/SurvBregDiv/reference/plot.cox_MDTL_enet.md) : Plot Validation Results for cox_MDTL_enet Object - [`plot(`*``*`)`](https://um-kevinhe.github.io/SurvBregDiv/reference/plot.cox_MDTL_ridge.md) : Plot Validation Results for cox_MDTL_ridge Object - [`plot(`*``*`)`](https://um-kevinhe.github.io/SurvBregDiv/reference/plot.StabSelect.md) : Plot Stability Selection Path - [`plot(`*``*`)`](https://um-kevinhe.github.io/SurvBregDiv/reference/plot.variable_importance.md) : Plot Variable Importance (Selection Frequency) - [`plot(`*``*`)`](https://um-kevinhe.github.io/SurvBregDiv/reference/plot.coxkl_enet.multi.md) : Plot Method for Multi-Source KL-Integrated Cox Elastic-Net Models ## Utilities - [`generate_eta()`](https://um-kevinhe.github.io/SurvBregDiv/reference/generate_eta.md) : Generate a Sequence of Tuning Parameters (eta) ## Datasets - [`ExampleData_lowdim`](https://um-kevinhe.github.io/SurvBregDiv/reference/ExampleData_lowdim.md) : Example low-dimensional survival data - [`ExampleData_highdim`](https://um-kevinhe.github.io/SurvBregDiv/reference/ExampleData_highdim.md) : Example high-dimensional survival data - [`ExampleData_cc_lowdim`](https://um-kevinhe.github.io/SurvBregDiv/reference/ExampleData_cc_lowdim.md) : Example Data for Conditional Logistic Regression - [`ExampleData_cc_highdim`](https://um-kevinhe.github.io/SurvBregDiv/reference/ExampleData_cc_highdim.md) : Example high-dimensional matched case-control data - [`ExampleData_indi`](https://um-kevinhe.github.io/SurvBregDiv/reference/ExampleData_indi.md) : Example internal/external Cox individual-level data - [`ExampleData_cc_indi`](https://um-kevinhe.github.io/SurvBregDiv/reference/ExampleData_cc_indi.md) : Example internal/external matched case-control individual-level data # Articles ### Get started - [SurvBregDiv: Bregman Divergence Data Integration for Time-to-Event Modelling](https://um-kevinhe.github.io/SurvBregDiv/articles/SurvBregDiv.md): ### Appendix - [Cross-validation Criteria](https://um-kevinhe.github.io/SurvBregDiv/articles/CV_Criteria.md): - [KL Divergence-Based Transfer Learning for Cox Model](https://um-kevinhe.github.io/SurvBregDiv/articles/coxkl.md): - [Cox KL Divergence: Handling Tied Event Times](https://um-kevinhe.github.io/SurvBregDiv/articles/coxkl_ties.md): - [NCC KL Divergence-Based Transfer Learning](https://um-kevinhe.github.io/SurvBregDiv/articles/ncc.md):