Comparing Time Series Boruta with Other Idionomic Methods for Identifying Treatment Processes in Experience Sampling Data from Adults with Trichotillomania

This study extends recently published research on GIMME-derived idionomic networks in adults with trichotillomania (Woolley et al., 2025; https://doi.org/10.1016/j.beth.2025.01.006). While GIMME effectively identifies nomothetic effects among interrelated variables, recent simulations indicate it may lack power to detect nonlinear effects that can be important for clinical decision-making (Li et al., 2025). We will complement previously known GIMME results with the findings from i-ARIMAX, i-Boruta, and TS-Boruta to identify processes that could improve clinical outcomes. The study will use experience sampling data from 54 individuals with trichotillomania who completed surveys over 30 days (average 70.5 surveys, SD=15.63; 78% completed at least 60 surveys). Building on Woolley et al.'s (2025) GIMME results, we will test whether i-ARIMAX and Boruta-based methods can provide additional clinically valuable insights. Along with the idiographic version of Boruta, we will implement a novel time series application of the Boruta machine learning algorithm in an idionomic framework. While past studies using idionomic methods have focused primarily on linear associations, we will test if TS-Boruta can identify important associations that linear methods may overlook. The study will advance our understanding of the utility of idionomic methods for process-based psychological treatment.