🤖 AI Summary
To address the clinical need for biomarkers that simultaneously enable disease screening (binary classification) and severity assessment (ordinal multi-class prediction), this paper proposes a novel multi-task learning framework that explicitly decouples these two heterogeneous yet interrelated tasks. Methodologically, we introduce a structured sparse regularization scheme—combining group Lasso and reweighted ℓ₁ penalties—to adaptively identify both shared and task-specific features. Furthermore, we integrate hierarchical modeling of ordinal responses to jointly capture the inherent ordinal structure and task heterogeneity. Extensive experiments on synthetic and real-world biomedical datasets demonstrate that our approach significantly improves predictive stability and generalization over conventional ordinal regression methods. Importantly, it yields interpretable biomarker selection and task-specific weight distributions, facilitating clinical interpretability and decision support.
📝 Abstract
The exploration of biomarkers, which are clinically useful biomolecules, and the development of prediction models using them are important problems in biomedical research. Biomarkers are widely used for disease screening, and some are related not only to the presence or absence of a disease but also to its severity. These biomarkers can be useful for prioritization of treatment and clinical decision-making. Considering a model helpful for both disease screening and severity prediction, this paper focuses on regression modeling for an ordinal response equipped with a hierarchical structure. If the response variable is a combination of the presence of disease and severity such as {{it healthy, mild, intermediate, severe}}, for example, the simplest method would be to apply the conventional ordinal regression model. However, the conventional model has flexibility issues and may not be suitable for the problems addressed in this paper, where the levels of the response variable might be heterogeneous. Therefore, this paper proposes a model assuming screening and severity prediction as different tasks, and an estimation method based on structural sparse regularization that leverages any common structure between the tasks when such commonality exists. In numerical experiments, the proposed method demonstrated stable performance across many scenarios compared to existing ordinal regression methods.
