Flexible tree-structured regression for clustered data with an application to quality of life in older adults

TL;DR

This paper introduces a flexible tree-structured regression method designed for clustered data, effectively modeling unit-specific effects and identifying subgroups, demonstrated through an application to quality of life in older adults.

Contribution

The paper presents a novel tree-structured approach that accounts for clustering and unit-specific effects, improving subgroup identification and variable selection in complex data.

Findings

Effective modeling of clustered data with unit-specific effects.

Successful application to quality of life data in older adults.

Competitive performance in variable selection and fit compared to alternatives.

Abstract

Tree-structured models are a powerful alternative to parametric regression models if non-linear effects and interactions are present in the data. Yet, classical tree-structured models might not be appropriate if data comes in clusters of units, which requires taking the dependence of observations into account. This is, for example, the case in cross-national studies, as presented here, where country-specific effects should not be neglected. To address this issue, we present a flexible tree-structured approach that achieves a sparse modeling of unit-specific effects and identifies subgroups (based on individual-level covariates) that differ with regard to the outcome. The methodological advances were motivated by the analysis of quality of life in older adults using data from the survey of Health, Ageing and Retirement in Europe. Application of the proposed model yields promising results and illustrated the accessibility of the approach. A comparison to alternative methods with regard to variable selection and goodness-of-fit was performed in several simulation experiments.