A machine-learning based causal mediation framework without the no-omitted-confounder assumption for latent variables and intensive longitudinal data

01/12/2023 to 30/11/2026

In this project a new framework for the analysis of mediator variables with latent variables is developed and extended to account for high-dimensional problems with many covariates as well as for intensive longitudinal data. The common basis in this framework is a method – the rank preserving model (RPM) – that can reliably identify mediators even if relevant confounder variables are omitted (Brandt, 2020; Ten Have et al., 2007; Zheng & Zhou, 2015). In the first part, a latent variable extension is proposed. In the second part, a machine learning based extension that accounts for high dimensional problems is developed.