Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing mortality prediction in cardiac arrest ICU patients through meta-modeling of structured clinical data from MIMIC-IV

Nursultan Mamatov, Philipp Kellmeyer

TL;DR

This study advances mortality risk prediction for ICU patients after cardiac arrest by fusing structured EHR variables with unstructured clinical notes from MIMIC-IV. It employs LASSO and XGBoost for robust feature selection and uses a transparent logistic regression meta-model, incorporating TF-IDF and BioBERT-derived text embeddings. The multimodal approach achieves a dramatic performance gain (AUC from ~0.75 to ~0.92) and superior clinical utility across a broad decision-curve range, underscoring the prognostic value of narrative notes. The work demonstrates that interpretable, multimodal models can enhance ICU risk stratification and supports extending this framework to external validation and real-time clinical decision support.

Abstract

Accurate early prediction of in-hospital mortality in intensive care units (ICUs) is essential for timely clinical intervention and efficient resource allocation. This study develops and evaluates machine learning models that integrate both structured clinical data and unstructured textual information, specifically discharge summaries and radiology reports, from the MIMIC-IV database. We used LASSO and XGBoost for feature selection, followed by a multivariate logistic regression trained on the top features identified by both models. Incorporating textual features using TF-IDF and BERT embeddings significantly improved predictive performance. The final logistic regression model, which combined structured and textual input, achieved an AUC of 0.918, compared to 0.753 when using structured data alone, a relative improvement 22%. The analysis of the decision curve demonstrated a superior standardized net benefit in a wide range of threshold probabilities (0.2-0.8), confirming the clinical utility of the model. These results underscore the added prognostic value of unstructured clinical notes and support their integration into interpretable feature-driven risk prediction models for ICU patients.

Enhancing mortality prediction in cardiac arrest ICU patients through meta-modeling of structured clinical data from MIMIC-IV

TL;DR

This study advances mortality risk prediction for ICU patients after cardiac arrest by fusing structured EHR variables with unstructured clinical notes from MIMIC-IV. It employs LASSO and XGBoost for robust feature selection and uses a transparent logistic regression meta-model, incorporating TF-IDF and BioBERT-derived text embeddings. The multimodal approach achieves a dramatic performance gain (AUC from ~0.75 to ~0.92) and superior clinical utility across a broad decision-curve range, underscoring the prognostic value of narrative notes. The work demonstrates that interpretable, multimodal models can enhance ICU risk stratification and supports extending this framework to external validation and real-time clinical decision support.

Abstract

Accurate early prediction of in-hospital mortality in intensive care units (ICUs) is essential for timely clinical intervention and efficient resource allocation. This study develops and evaluates machine learning models that integrate both structured clinical data and unstructured textual information, specifically discharge summaries and radiology reports, from the MIMIC-IV database. We used LASSO and XGBoost for feature selection, followed by a multivariate logistic regression trained on the top features identified by both models. Incorporating textual features using TF-IDF and BERT embeddings significantly improved predictive performance. The final logistic regression model, which combined structured and textual input, achieved an AUC of 0.918, compared to 0.753 when using structured data alone, a relative improvement 22%. The analysis of the decision curve demonstrated a superior standardized net benefit in a wide range of threshold probabilities (0.2-0.8), confirming the clinical utility of the model. These results underscore the added prognostic value of unstructured clinical notes and support their integration into interpretable feature-driven risk prediction models for ICU patients.
Paper Structure (47 sections, 5 equations, 13 figures, 9 tables)

This paper contains 47 sections, 5 equations, 13 figures, 9 tables.

Table of Contents

  1. Abstract
  2. Introduction
  3. Background and Motivation
  4. Problem Statement
  5. Objectives and Research Questions
  6. Significance and Contribution
  7. Literature Review
  8. Overview
  9. ICU Mortality Prediction Using Structured Data
  10. Text-Based Modeling and ClinicalBERT
  11. Multimodal Learning and Data Integration
  12. Clinical Context: Post–Cardiac Arrest Outcomes
  13. Interpretability and Responsible AI
  14. Summary
  15. Data Description
  16. ...and 32 more sections

Figures (13)

  • Figure 1: Simplified flow of data capture across MIMIC-IV modules, adapted from harutyunyan2019multitask.
  • Figure 2: ROC curve comparison after adding textual features.
  • Figure 3: Decision curve analysis comparing structured-only vs. multimodal models.
  • Figure 4: Distribution of Vital Signs across the first 24 hours of ICU admission.
  • Figure 5: Distributions of Lab Test Values
  • ...and 8 more figures