Deep Learning-Based Detection of the Acute Respiratory Distress Syndrome: What Are the Models Learning?

TL;DR

This study demonstrates that deep neural networks using ventilator waveform data can outperform traditional models in detecting ARDS, revealing that high-frequency physiologic signals contribute significantly to model performance and interpretability.

Contribution

The paper introduces a CNN-based ARDS detection model that surpasses prior models and explores the importance of high-frequency data in physiologic signals for deep learning interpretability.

Findings

CNN model outperforms random forest in ARDS detection

High-frequency physiologic data improves model performance

Model learns features from both low and high frequency domains

Abstract

The acute respiratory distress syndrome (ARDS) is a severe form of hypoxemic respiratory failure with in-hospital mortality of 35-46%. High mortality is thought to be related in part to challenges in making a prompt diagnosis, which may in turn delay implementation of evidence-based therapies. A deep neural network (DNN) algorithm utilizing unbiased ventilator waveform data (VWD) may help to improve screening for ARDS. We first show that a convolutional neural network-based ARDS detection model can outperform prior work with random forest models in AUC (0.95+/-0.019 vs. 0.88+/-0.064), accuracy (0.84+/-0.026 vs 0.80+/-0.078), and specificity (0.81+/-0.06 vs 0.71+/-0.089). Frequency ablation studies imply that our model can learn features from low frequency domains typically used for expert feature engineering, and high-frequency information that may be difficult to manually featurize. Further experiments suggest that subtle, high-frequency components of physiologic signals may explain the superior performance of DL models over traditional ML when using physiologic waveform data. Our observations may enable improved interpretability of DL-based physiologic models and may improve the understanding of how high-frequency information in physiologic data impacts the performance our DL model.