Variable Screening Methods in Conditional Logistic Individual Level Models of Disease Spread

Abstract

The conditional logistic individual-level model is a recently developed infectious disease model, particularly suited for modeling spatial-based infection risk. It is designed to reduce computational complexity and expand the range of available statistical software for data analysis (Akter & Deardon, 2025). This study aims to apply and evaluate different variable selection techniques for the newly introduced conditional logistic individual-level models (CL-ILMs). These variable selection methods include forward and backward stepwise Akaike information criterion (AIC), least absolute shrinkage and selection operator (Lasso), spike-and-slab prior (SS prior), and two-stage screening methods. The ultimate goal is to boost model performance and interpretability, and to reduce the risk of overfitting ultimately leading to more robust and effective models. We examine and compare the performance of these methods using simulated data and real-life data from the outbreak of foot-and-mouth disease in the UK in 2001.