Temporal Sepsis Modeling: a Fully Interpretable Relational Way

2026-01-29

AI Summary

This study addresses the limitations of insufficient model interpretability and neglect of patient subtypes in early sepsis prediction by proposing a novel modeling paradigm grounded in relational data perspectives. The approach treats longitudinal electronic health record data as relational multivariate logs and employs propositionalization techniques—specifically aggregation and selection functions—to generate inherently interpretable features. A selective naive Bayes classifier is then applied for prediction. The method achieves competitive predictive performance while offering four complementary forms of interpretability: univariate, global, local, and counterfactual explanations. This multi-faceted interpretability significantly enhances the transparency and trustworthiness of clinical decision support, enabling clinicians to better understand and act upon model predictions in critical care settings.

Technology Category

• Reasoning under Uncertainty: Relational Probabilistic Models
• Machine Learning: Transparent, Interpretable, Explainable ML
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• Security and Privacy: Data transparency and provenance

Abstract

Sepsis remains one of the most complex and heterogeneous syndromes in intensive care, characterized by diverse physiological trajectories and variable responses to treatment. While deep learning models perform well in the early prediction of sepsis, they often lack interpretability and ignore latent patient sub-phenotypes. In this work, we propose a machine learning framework by opening up a new avenue for addressing this issue: a relational approach. Temporal data from electronic medical records (EMRs) are viewed as multivariate patient logs and represented in a relational data schema. Then, a propositionalisation technique (based on classic aggregation/selection functions from the field of relational data) is applied to construct interpretable features to"flatten"the data. Finally, the flattened data is classified using a selective naive Bayesian classifier. Experimental validation demonstrates the relevance of the suggested approach as well as its extreme interpretability. The interpretation is fourfold: univariate, global, local, and counterfactual.

Problem

• sepsis
• interpretability
• patient sub-phenotypes
• temporal modeling
• heterogeneous syndrome

Innovation

• relational modeling
• propositionalisation
• interpretable machine learning
• temporal sepsis prediction
• selective naive Bayes