ISeeU2: Visually Interpretable ICU mortality prediction using deep learning and free-text medical notes (2020-05-19T00:00:00.000000Z)

TL;DR

A Deep Learning model trained on MIMIC-III is shown to predict mortality using raw nursing notes, together with visual explanations for word importance, outperforming the traditional SAPS-II score and providing enhanced interpretability when compared with similar Deep Learning approaches.

Abstract

Accurate mortality prediction allows Intensive Care Units (ICUs) to adequately benchmark clinical practice and identify patients with unexpected outcomes. Traditionally, simple statistical models have been used to assess patient death risk, many times with sub-optimal performance. On the other hand deep learning holds promise to positively impact clinical practice by leveraging medical data to assist diagnosis and prediction, including mortality prediction. However, as the question of whether powerful Deep Learning models attend correlations backed by sound medical knowledge when generating predictions remains open, additional interpretability tools are needed to foster trust and encourage the use of AI by clinicians. In this work we show a Deep Learning model trained on MIMIC-III to predict mortality using raw nursing notes, together with visual explanations for word importance. Our model reaches a ROC of 0.8629 (+/-0.0058), outperforming the traditional SAPS-II score and providing enhanced interpretability when compared with similar Deep Learning approaches.

Authors

William Caicedo-Torres

1 papers

Jairo Gutiérrez

1 papers

TL;DR

Abstract

Authors

References37 items

Deep Learning Models for Detecting Malware Attacks

Fair Allocation of Scarce Medical Resources in the Time of Covid-19.

Critical Care Utilization for the COVID-19 Outbreak in Lombardy, Italy: Early Experience and Forecast During an Emergency Response.

ISeeU: Visually interpretable deep learning for mortality prediction inside the ICU

Improving Hospital Mortality Prediction with Medical Named Entities and Multimodal Learning

Patient representation learning and interpretable evaluation using clinical notes

Benchmarking deep learning models on large healthcare datasets

The MIMIC Code Repository: enabling reproducibility in critical care research

Combining LSTM and Latent Topic Modeling for Mortality Prediction

Deep Learning for Medical Image Analysis

Deep EHR: A Survey of Recent Advances in Deep Learning Techniques for Electronic Health Record (EHR) Analysis

A Unified Approach to Interpreting Model Predictions

Learning Important Features Through Propagating Activation Differences

Neural Document Embeddings for Intensive Care Patient Mortality Prediction

Model-Agnostic Interpretability of Machine Learning

Directly Modeling Missing Data in Sequences with RNNs: Improved Classification of Clinical Time Series

The mythos of model interpretability

Recurrent Neural Networks for Multivariate Time Series with Missing Values

MIMIC-III, a freely accessible critical care database

TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems

Temporal Convolutional Neural Networks for Diagnosis from Lab Tests

Adam: A Method for Stochastic Optimization

Striving for Simplicity: The All Convolutional Net

Scoring systems in the intensive care unit: A compendium

Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps

An Efficient Explanation of Individual Classifications using Game Theory

Long Short-Term Memory

A new Simplified Acute Physiology Score (SAPS II) based on a European/North American multicenter study

Artificial intelligence in medicine

A Value for n-person Games

Deep Patient Representation of Clinical Notes via Multi-Task Learning for Mortality Prediction.

Kass-hout, Improving Hospital Mortality Prediction with

Deep Learning

Gradient-based learning applied to document recognition

Haffner

A new Simplified Acute Physiology Score (SAPS II) based on a European/North American multicenter study.

Fair Allocation of 26

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names