Learning to Diagnose with LSTM Recurrent Neural Networks
By Zachary Chase Lipton, David C. Kale, Charles Peter Elkan, Randall C. Wetzel. 2015
1,224 citations recorded by the source corpus.
Clinical medical data, especially in the intensive care unit (ICU), consist of multivariate time series of observations. For each patient visit (or episode), sensor data and lab test results are recorded in the patient's Electronic Health Record (EHR). While potentially containing a wealth of insights, the data is difficult to mine effectively, owing to varying length, irregular sampling and missing data. Recurrent Neural Networks (RNNs), particularly those using Long Short-Term Memory (LSTM) hidden units, are powerful and increasingly popular models for learning from sequence data. They effectively model varying length sequences and capture long range dependencies. We present the first study to empirically evaluate the ability of LSTMs to recognize patterns in multivariate time series of clinical measurements. Specifically, we consider multilabel classification of diagnoses, training a model to classify 128 diagnoses given 13 frequently but irregularly sampled clinical measurements. First, we establish the effectiveness of a simple LSTM network for modeling clinical data. Then we demonstrate a straightforward and effective training strategy in which we replicate targets at each sequence step. Trained only on raw time series, our models outperform several strong baselines, including a multilayer perceptron trained on hand-engineered features.
Subfields: Artificial Intelligence, Model Architectures, Medical & Health AI, Signal Processing
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Mohit Bansal
- Jianfeng Gao
- Yoshua Bengio
- Percy Liang
- Quoc V. Le
- Luke S. Zettlemoyer
- Ruslan Salakhutdinov
- Yejin Choi
In Model Architectures
- Mihaela van der Schaar
- Wei Zhang
- Jimeng Sun
- Edward Choi
- Bryan Lim
- Quoc V. Le
- Hanxiao Liu
- Zachary Chase Lipton
In Medical & Health AI
- Mihaela van der Schaar
- Jimeng Sun
- Cao Xiao
- Bryan Lim
- Edward Choi
- Fei Wang
- Joseph D. Futoma
- Hongyuan Zha
In Signal Processing
- Zachary Chase Lipton
- David C. Kale
- Randall C. Wetzel
- Bryan Lim
- Yuan Luo
- Edward Choi
- Mihaela van der Schaar
- Jimeng Sun
Citations in the corpus
- Deep Learning for IoT Big Data and Streaming Analytics: A Survey (2017, background): “[133] investigated the performance of LSTM networks to analyze and recognize patterns in multivariate time series of medical measurements in intensive care units (ICUs).”
- RETAIN: An Interpretable Predictive Model for Healthcare using Reverse Time Attention Mechanism (2016, methodology): “In case of learning to diagnose (L2D) [30], the input vector xi consists of continuous clinical measures.”
- Distilling Knowledge from Deep Networks with Applications to Healthcare Domain (2015, background): “With the flexibility and power of all kinds of neural networks, some deep network variants are also potentially suitable for healthcare tasks [9, 30].”
- Reinforcement Learning in Healthcare: A Survey (2019, background): “Normally formulated as a supervised classification problem, existing machining learning methods on clinical diagnosis heavily rely on a large number of annotated samples in order to infer and predict the possible diagnoses [244], [245], [246].”
- Forecasting Across Time Series Databases using Long Short-Term Memory Networks on Groups of Similar Series (2017, background): “…Natural Language Processing (Mikolov et al., 2010), machine translation (Sutskever et al., 2014), and speech recognition (Graves et al., 2013), and are also gaining popularity in time series research (Fei and Yeung, 2015; Paw lowski and Kurach, 2015; Lipton et al., 2015; Zimmermann et al., 2012).”
- Interpretable Representation Learning for Healthcare via Capturing Disease Progression through Time (2018, background): “Several recent studies showed that predictive modeling of EHR data using RNNs can significantly outperform traditional models such as logistic regression, multilayer perceptron (MLP), and support vector machine (SVM), which all depend on feature engineering [6, 18, 23].”
- A Multi-Horizon Quantile Recurrent Forecaster (2017, background): “The benefit behind forking-sequences may also be related to ideas described in Lipton et al, 2015, where a scalar categorical target is replicated to each recurrent layer in a time series classification problem.”
- Natural Language Processing for EHR-Based Computational Phenotyping (2018, methodology): “[163] evaluated the performance of LSTM in phenotype prediction using multivariate time series clinical measurements.”
- Recurrent Neural Networks for Classifying Relations in Clinical Notes (2017, neutral): “[41] trained Long Short-Term Memory networks (LSTMs) to classify 128 diagnoses from 13 frequently but irregularly sampled clinical measurements from patients in pediatric ICU.”
- Recurrent Neural Networks for Multivariate Time Series with Missing Values (2016, critical): “Recently RNN models have been explored for modeling diseases and patient diagnosis in health care domain [14, 5, 19] using doctor notes but are not readily applicable for comparison in our time series classification tasks since they don’t handle missing data.”
- Scalable and accurate deep learning with electronic health records (2018, background): “nformation Mart for Intensive Care (MIMIC) data [48], many prior studies also have focused on ICU patients from a single center [92, 41]; other single-center studies have also focused on ICU patients [61]. Each ICU patient has significantly more data available than each general hospital patient, alth”
- Review of Bridge Structural Health Monitoring Aided by Big Data and Artificial Intelligence: From Condition Assessment to Damage Detection (2020, neutral): “In classification applications, RNNs are appropriate models for mining variable-length time series with irregularly sampled or missing measurements, for example, using LSTM to classify clinical monitoring data with bad data condition for diagnosis (Lipton et al. 2015).”
- Deep EHR: A Survey of Recent Advances in Deep Learning Techniques for Electronic Health Record (EHR) Analysis (2017, neutral): “[43], [45] [38], [44], [54] [52], [53], [60], [61] [13], [14]”
- Shifting machine learning for healthcare from development to deployment and from models to data (2022, background): “In particular, long short-term memory networks 109 and gated units 110 were staple RNNs in modelling EHR data, as these networks can model the sequential nature of clinical data 111,112 and clinical text 5,104,105,113 .”