surtvep

surtvep estimates time-varying effects in Cox non-proportional hazards models for large-scale time-to-event data.

Key Features:

• Computational efficiency: Implements a Kronecker product-based proximal algorithm to address memory limitations and numerical instability with large datasets.
• Model support: Supports both unstratified and stratified Cox non-proportional hazards models.
• Penalized estimation: Incorporates P-spline and smoothing spline penalties to improve estimation of time-varying effects.
• Tuning parameter selection: Provides cross-validation and information criteria for determining optimal tuning parameters.
• Parallel computation: Enables parallel computation to enhance computational efficiency.
• Operating characteristics: Provides estimated time-varying effects, confidence intervals, hypothesis testing, and estimated hazard functions and survival probabilities.

Scientific Applications:

• Medical studies with registry data: Analyzes large-scale time-to-event data from national disease registries where detecting and accounting for time-varying effects is crucial.
• Survival analysis with dynamic effect trajectories: Models scenarios where covariate effects change over the follow-up period, particularly when standard proportional hazards assumptions are violated.

Methodology:

surtvep implements a computationally efficient Kronecker product-based proximal algorithm with P-spline and smoothing spline penalties for estimating time-varying effects in Cox non-proportional hazards models.