Penalized DLNM Framework

What problem does it solve?

This Skill addresses the challenge of selecting appropriate smoothing parameters for Distributed Lag Non-Linear Models (DLNMs) by automating the process using penalized splines and Generalized Additive Models (GAMs).

Core Features & Use Cases

• Automatic Smoothing: Leverages mgcv::gam() and cbPen() to estimate optimal smoothing parameters from data, removing the need for manual selection.
• Flexible Model Fitting: Integrates penalized cross-basis terms with other smooth terms (e.g., s(date)) within a GAM framework.
• Use Case: When exploring exposure-lag-response relationships where the optimal degree of smoothness is uncertain, this Skill provides a data-driven approach to model fitting, ensuring more robust and reproducible results.

Quick Start

Use the Penalized DLNM Framework skill to fit a GAM model with penalized cross-basis terms for exposure and date.