Transparent Early ICU Mortality Prediction with Clinical Transformer and Per-Case Modality Attribution

TL;DR

This paper introduces a transparent, multimodal ensemble model for early ICU mortality prediction that combines physiological data and clinical notes, providing interpretability and robustness for clinical use.

Contribution

It presents a novel, interpretable ensemble architecture that fuses time-series and unstructured clinical notes with per-case attribution, improving predictive performance and transparency.

Findings

Improved AUPRC and AUROC over single models on MIMIC-III

Maintains calibration and robustness with missing modalities

Provides multilevel interpretability and per-case attribution

Abstract

Early identification of intensive care patients at risk of in-hospital mortality enables timely intervention and efficient resource allocation. Despite high predictive performance, existing machine learning approaches lack transparency and robustness, limiting clinical adoption. We present a lightweight, transparent multimodal ensemble that fuses physiological time-series measurements with unstructured clinical notes from the first 48 hours of an ICU stay. A logistic regression model combines predictions from two modality-specific models: a bidirectional LSTM for vitals and a finetuned ClinicalModernBERT transformer for notes. This traceable architecture allows for multilevel interpretability: feature attributions within each modality and direct per-case modality attributions quantifying how vitals and notes influence each decision. On the MIMIC-III benchmark, our late-fusion ensemble improves discrimination over the best single model (AUPRC 0.565 vs. 0.526; AUROC 0.891 vs. 0.876) while maintaining well-calibrated predictions. The system remains robust through a calibrated fallback when a modality is missing. These results demonstrate competitive performance with reliable, auditable risk estimates and transparent, predictable operation, which together are crucial for clinical use.