Authors: Brandon Silva, Miguel Contreras, Sabyasachi Bandyopadhyay, Yuanfang Ren, Ziyuan Guan, Jeremy Balch, Kia Khezeli, Tezcan Ozrazgat Baslanti, Ben Shickel, Azra Bihorac, Parisa Rashidi

Published on: March 11, 2024

Impact Score: 8.0

Arxiv code: Arxiv:2403.07201

Summary

• What is new: The research offers a dynamic prediction model for acute brain dysfunction (ABD) including delirium, coma, and mortality in ICU patients on a 12-hour interval basis, using large-scale public datasets.
• Why this is important: Current ABD diagnostic methods in the ICU rely on infrequent clinical observations, leading to delayed intervention.
• What the research proposes: Developed automated methods using EHR data to predict ABD states dynamically throughout an ICU stay, utilizing two advanced neural network models.
• Results: Achieved a high predictive performance with a mean AUROC of 0.95 for predicting ABD outcomes and 0.79 for transitions between ABD states.

Technical Details

Technological frameworks used: MAMBA selective state space, Longformer Transformer

Models used: Neural network models

Data used: Large publicly available datasets including MIMIC-IV and eICU

Potential Impact

Healthcare providers, ICU technology developers, medical diagnostic companies

Want to implement this idea in a business?

We have generated a startup concept here: NeuroPredict.