boyiguo1 / manuscript-bh_additive_cox Goto Github PK
View Code? Open in Web Editor NEWRepo to replicate the manuscript Bayesian Hierarchical Additive Cox Model
Home Page: https://arxiv.org/abs/2205.11600
Repo to replicate the manuscript Bayesian Hierarchical Additive Cox Model
Home Page: https://arxiv.org/abs/2205.11600
The article is about additive hazards models, and hence not read thoroughly.
The article extends the trend filtering additive model to the Cox proportional hazard diagram. Trend filtering additive functions are piece-wise polynomials with adaptively chosen knots. The likelihood function consists of a trend filtering penalty for knots selection and a group lasso penalty for functional selection. A second tuning parameter,
$\alpha$ , is used to control the tradeoff between two penalties. Model fitting is achieved by the proximal gradient descent algorithm and hence is capable of high-dimensional data analysis. An R package tfCox is developed. One of the problems for this approach is that the estimated curve is not smooth, or even continuous depending on the choice of trend filtering funciton.
The article is about multivariate functional predictors and doesn't apply to high-dimensional data analysis. A good place to find other "low-dimensional" additive Cox models. Could be used in the discussion section
The author demonstrated using generalized additive model infrastructure to model time-to-event outcome via piece-wise exponential model. This modeling approach increases the utility, allowing spatial(-temporal) random effects and time-dependent covariates. The authors also introduced a utility R-package that does pre-process step of data before fitting to a generalized model. The article doesn't directly relate to the current work as not in the Cox paradigm, but supplies arguments for introduction and discussion.
The authors propose a flexible survival additive models that is link-based to model time-to-event with mixed censoring. Different link functions can be used to achieve different interpretations of the beta coefficients. While the model provides alternative utility for modeling time-to-event outcomes in comparison to the Cox model or Accelerated failure time model, high-dimensionality of the predictors could be addressed under this framework as sparsity penalty was not imposed. An R package GJRM is provided.
Contents include:
offset induce probelm
Read the Simulation Functions of BhGLM to understand what Dr. Yi meant abt effect variance decomposition.
Adjust the simulation error, such that the baseline AUC (mgcv model result) is around 0.75
Examine main effects
Examine interaction effects
Current Status: Going through second round of revision
End goal: to send to the co-authors to review before 08/01/2022
This is not intuitive as mentioned by Dr. Rahman.
Possibly conducting a new simulation with Just LASSO model
Create tables and viz for both measures using targets
static branching (764b823)
Include the results in the manuscript
Document the result
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.