Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Realtime, multimodal invasive ventilation risk monitoring using language models and BoXHED

Arash Pakbin, Aaron Su, Donald K. K. Lee, Bobak J. Mortazavi

TL;DR

This study proposes an innovative approach to enhance iV risk monitoring by incorporating clinical notes into the monitoring pipeline through using language models for text summarization, paving the way for improved patient care and informed clinical decision-making in ICU settings.

Abstract

Objective: realtime monitoring of invasive ventilation (iV) in intensive care units (ICUs) plays a crucial role in ensuring prompt interventions and better patient outcomes. However, conventional methods often overlook valuable insights embedded within clinical notes, relying solely on tabular data. In this study, we propose an innovative approach to enhance iV risk monitoring by incorporating clinical notes into the monitoring pipeline through using language models for text summarization. Results: We achieve superior performance in all metrics reported by the state-of-the-art in iV risk monitoring, namely: an AUROC of 0.86, an AUC-PR of 0.35, and an AUCt of up to 0.86. We also demonstrate that our methodology allows for more lead time in flagging iV for certain time buckets. Conclusion: Our study underscores the potential of integrating clinical notes and language models into realtime iV risk monitoring, paving the way for improved patient care and informed clinical decision-making in ICU settings.

Realtime, multimodal invasive ventilation risk monitoring using language models and BoXHED

TL;DR

This study proposes an innovative approach to enhance iV risk monitoring by incorporating clinical notes into the monitoring pipeline through using language models for text summarization, paving the way for improved patient care and informed clinical decision-making in ICU settings.

Abstract

Objective: realtime monitoring of invasive ventilation (iV) in intensive care units (ICUs) plays a crucial role in ensuring prompt interventions and better patient outcomes. However, conventional methods often overlook valuable insights embedded within clinical notes, relying solely on tabular data. In this study, we propose an innovative approach to enhance iV risk monitoring by incorporating clinical notes into the monitoring pipeline through using language models for text summarization. Results: We achieve superior performance in all metrics reported by the state-of-the-art in iV risk monitoring, namely: an AUROC of 0.86, an AUC-PR of 0.35, and an AUCt of up to 0.86. We also demonstrate that our methodology allows for more lead time in flagging iV for certain time buckets. Conclusion: Our study underscores the potential of integrating clinical notes and language models into realtime iV risk monitoring, paving the way for improved patient care and informed clinical decision-making in ICU settings.
Paper Structure (12 sections, 2 equations, 6 figures, 1 table)

This paper contains 12 sections, 2 equations, 6 figures, 1 table.

Table of Contents

  1. INTRODUCTION
  2. RELATED WORKS
  3. RESULTS
  4. Cohort Selection Process
  5. Finding the Best Note Embedding Dimensionality
  6. Realtime iV Risk Monitoring: Comparing against State-of-the-art
  7. DISCUSSION
  8. CONCLUSION
  9. MATERIALS AND METHODS
  10. Numerically Representing Notes using Language Models
  11. Tracking Realtime Risk on Fused Data
  12. Evaluating Realtime Risk

Figures (6)

  • Figure 1: The overall structure of BoXHEDMM. It integrates tabular time-series measurements with information extracted from the most recent clinical note. The combined information feed into our realtime risk monitoring system.
  • Figure 2: ROC and PR plots comparing BoXHEDMM with the approach from pakbin2023predicting for iV prognostication. AUROC and AUC-PR values are also provided, where higher values indicate better performance for both metrics.
  • Figure 3: AUCt values with 95% confidence intervals for iV prognostication. A higher AUCt value indicates better concordance.
  • Figure 4: Histogram of lead times for flagging the predicted cases of iV.
  • Figure 5: Comparison of relative variable importances for BoXHEDMM and the methodology from pakbin2023predicting. The plot includes only those variables with relative importances greater than $0.1$. Highlighted bars correspond to notes embeddings.
  • ...and 1 more figures