EF-Net: A Deep Learning Approach Combining Word Embeddings and Feature Fusion for Patient Disposition Analysis

TL;DR

The study tackles emergency department overcrowding by predicting multiclass patient disposition using EF-Net, a neural network that embeds categorical features and fuses them with numerical features. It combines EF-Net with XGBoost through soft voting, yielding the highest performance on the MIMIC-IV-ED v2.0 dataset (accuracy up to 96% for the ensemble). EF-Net outperforms a baseline neural network and prior work in accuracy, AUROC, precision, recall, and F1, demonstrating the value of semantic categorical embeddings and feature fusion for ED decision support. While promising, the work relies on single-institution data and omits laboratory/imaging data, suggesting avenues for broader validation and richer feature integration. The accompanying code is publicly available, enabling replication and extension of the approach in clinical settings.

Abstract

One of the most urgent problems is the overcrowding in emergency departments (EDs), caused by an aging population and rising healthcare costs. Patient dispositions have become more complex as a result of the strain on hospital infrastructure and the scarcity of medical resources. Individuals with more dangerous health issues should be prioritized in the emergency room. Thus, our research aims to develop a prediction model for patient disposition using EF-Net. This model will incorporate categorical features into the neural network layer and add numerical features with the embedded categorical features. We combine the EF-Net and XGBoost models to attain higher accuracy in our results. The result is generated using the soft voting technique. In EF-Net, we attained an accuracy of 95.33%, whereas in the Ensemble Model, we achieved an accuracy of 96%. The experiment's analysis shows that EF-Net surpasses existing works in accuracy, AUROC, and F1-Score on the MIMIC-IV-ED dataset, demonstrating its potential as a scalable solution for patient disposition assessment. Our code is available at https://github.com/nafisa67/thesis

Figures (5)

• Figure 1: Distribution of Disposition Classes in MIMIC-IV ED dataset.
• Figure 2: Flowchart of Proposed Methodology.
• Figure 3: Architecture of EF-Net Model.
• Figure 4: Comparison of confusion matrices for EF-Net and Ensemble Model.
• Figure 5: Training vs Validation Loss. Early stopping criteria is used to avoid overfitting of the model.