Auditing Construct Overlap in Explainable Machine Learning: Evidence from Burnout-Depression Prediction Across Student Cohorts

2026-07-12

AI Summary

This study addresses a critical pitfall in explainable machine learning when predicting composite mental health constructs such as burnout-depression: overlapping latent constructs can induce spuriously stable feature importance rankings, thereby misleading assessments of model generalizability. To mitigate this, the authors propose a transferable residualization testing protocol that integrates ElasticNet regression, Kendall’s τ rank consistency analysis, and cross-cohort validation to systematically disentangle shared variance between predictors and outcomes. Empirical results reveal a high correlation (r = 0.72) between trait anxiety and depressive symptom subscales; after residualization, model R² plummeted from 0.41 to as low as 0.016, with prediction intervals spanning 35.4 points, indicating unsuitability for individual-level inference. This approach provides a crucial auditing tool for XAI research, demonstrating that apparent stability often stems from construct confounding rather than genuine predictive signal.

Abstract

Explainable machine learning (XML) pipelines applied to composite mental health outcomes can produce apparently-robust, cross-population-stable risk hierarchies that are largely artefacts of how the outcome was constructed. We demonstrate this using an ElasticNet pipeline applied to 886 medical students at the University of Lausanne (primary cohort, 2022), validated across 2,580 longitudinal observations at three time points and 701 non-medical students from eight faculties; all three datasets share identical instruments. The pipeline produces a hierarchy in which trait anxiety and health satisfaction dominate wherever the outcome is measured, with Kendall $τ= 1.0$ for the top-two positions across all five evaluation sets and consistent transfer performance ($R^2$: 0.41-0.49). Two residualization experiments, which isolate shared variance between correlated variables via regression, reveal the mechanism: when trait anxiety (STAI-T) is residualized against the co-included depression subscale (CES-D, $r = 0.72$), model $R^2$ drops from 0.41 to 0.16 and STAI-T falls from rank 1 to rank 6; when burnout subscales are residualized against CES-D, $R^2$ collapses to 0.016. Prediction intervals average 35.4 units on a 0-100 scale (2.4 outcome standard deviations), independently ruling out individual-level deployment. The residualization protocol is the paper's transferable contribution: any XAI study combining correlated predictor and outcome constructs should apply this check before interpreting apparent stability as a finding.

Problem

• explainable machine learning
• construct overlap
• mental health prediction
• burnout-depression
• outcome construction

Innovation

• construct overlap
• residualization protocol
• explainable machine learning
• mental health prediction
• feature importance stability