• FluTrackers.com Inc. does not provide medical advice. Information on this web site is collected from various internet resources, and the FluTrackers board of directors makes no warranty to the safety, efficacy, correctness or completeness of the information posted on this site by any author or poster. The information collated here is for instructional and/or discussion purposes only and is NOT intended to diagnose or treat any disease, illness, or other medical condition. Every individual reader or poster should seek advice from their personal physician/healthcare practitioner before considering or using any interventions that are discussed on this website. By continuing to access this website you agree to consult your personal physican before using any interventions posted on this website, and you agree to hold harmless FluTrackers.com Inc., the board of directors, the members, and all authors and posters for any effects from use of any medication, supplement, vitamin or other substance, device, intervention, etc. mentioned in posts on this website, or other internet venues referenced in posts on this website.
  • We are not asking for any donations. Do not donate to any entity who says they are raising funds for us.

PLoS One . Hybrid feature-selection and diversity-guided stacking framework for interpretable ensemble learning: Application to COVID-19 mortality p

tetano

Editor, Senior Moderator
PLoS One


. 2026 Apr 17;21(4):e0341198.
doi: 10.1371/journal.pone.0341198. eCollection 2026.
Hybrid feature-selection and diversity-guided stacking framework for interpretable ensemble learning: Application to COVID-19 mortality prediction

Farideh Mohtasham[SUP] 1 [/SUP], Seyed Saeed Hashemi Nazari[SUP] 2 [/SUP], Mohamad Amin Pourhoseingholi[SUP] 3 [/SUP], Kaveh Kavousi[SUP] 4 [/SUP], Mohammad Reza Zali[SUP] 1 [/SUP]


Affiliations
Abstract

Background: Reliable predictive modeling in high-dimensional biomedical data requires a balance between accuracy, interpretability, and computational efficiency. However, existing ensemble methods often overlook model diversity or rely on ad hoc feature-selection approaches, which limit generalizability. This study introduces a hybrid feature-selection and diversity-guided stacking framework designed to improve robustness and scalability across clinical and other data-intensive domains.
Methods: The proposed framework integrates a hybrid feature-selection pipeline-combining Variance Inflation Factor (VIF), Analysis of Variance (ANOVA), Sequential Backward Elimination (SBE), and Lasso regression-to reduce multicollinearity and overfitting. It also employs a diversity-aware stacking strategy that constructs sub-model sets based on pairwise diversity measures (Disagreement, Yule's Q, and Cohen's Kappa) and non-pairwise metrics (Entropy and Kohavi-Wolpert). Sixteen base classifiers and five meta-learners were trained using repeated 10-fold cross-validation. The framework was evaluated using data from 4,778 hospitalized COVID-19 patients with 116 clinical and laboratory attributes, preprocessed using robust scaling and ROSE-based class balancing.
Results: The optimal configuration, which stacked Random Forest and XGBoost models using a Neural Network meta-learner, achieved 91.4% accuracy (95% CI: 89.8-92.8), AUC = 0.955, F1 = 0.801, and MCC = 0.746, outperforming the best individual model (AdaBoost, 90.2%). Training time (~450 s) and per-case inference time (<0.2 s) demonstrated computational feasibility. Feature-importance analysis and SHAP-based interpretation confirmed clinical relevance and interpretability.
Conclusions: The hybrid feature-selection and diversity-guided stacking framework improves predictive accuracy and interpretability while maintaining computational efficiency. Although validated using COVID-19 mortality data, the approach is broadly applicable to biomedical, environmental, and engineering prediction tasks that require interpretable and scalable ensemble learning.


 
Back
Top Bottom