Modeling Missing Data in Clinical Time Series with RNNs

Abstract

We demonstrate a simple strategy to cope with missing data in sequential inputs, addressing the task of multilabel classification of diagnoses given clinical time series. Collected from the pediatric intensive care unit (PICU) at Children's Hospital Los Angeles, our data consists of multivariate time series of observations. The measurements are irregularly spaced, leading to missingness patterns in temporally discretized sequences. While these artifacts are typically handled by imputation, we achieve superior predictive performance by treating the artifacts as features. Unlike linear models, recurrent neural networks can realize this improvement using only simple binary indicators of missingness. For linear models, we show an alternative strategy to capture this signal. Training models on missingness patterns only, we show that for some diseases, what tests are run can be as predictive as the results themselves.

References in corpus (1)

• A Critical Review of Recurrent Neural Networks for Sequence Learning

Cited by in corpus (37)

• A Survey on the Explainability of Supervised Machine Learning
• Recurrent Neural Networks for Multivariate Time Series with Missing Values
• Explainable Artificial Intelligence: A Survey of Needs, Techniques, Applications, and Future Direction
• Training recurrent neural networks robust to incomplete data: application to Alzheimer's disease progression modeling
• Dynamic Spatiotemporal Graph Convolutional Neural Networks for Traffic Data Imputation with Complex Missing Patterns
• MGP-AttTCN: An Interpretable Machine Learning Model for the Prediction of Sepsis
• Attend and Diagnose: Clinical Time Series Analysis using Attention Models
• Dropout Feature Ranking for Deep Learning Models
• Deepr: A Convolutional Net for Medical Records
• Partially-Supervised Image Captioning
• "The Human Body is a Black Box": Supporting Clinical Decision-Making with Deep Learning
• Why Not to Use Zero Imputation? Correcting Sparsity Bias in Training Neural Networks
• Temporal-Clustering Invariance in Irregular Healthcare Time Series
• Deep Learning to Attend to Risk in ICU
• Effectiveness of LSTMs in Predicting Congestive Heart Failure Onset
• Are deep learning models superior for missing data imputation in large surveys? Evidence from an empirical comparison
• Medical Diagnosis From Laboratory Tests by Combining Generative and Discriminative Learning
• Learning to Identify Patients at Risk of Uncontrolled Hypertension Using Electronic Health Records Data
• Learning Hierarchical Representations of Electronic Health Records for Clinical Outcome Prediction
• Learning Representations of Missing Data for Predicting Patient Outcomes
• Artificial neural networks condensation: A strategy to facilitate adaption of machine learning in medical settings by reducing computational burden
• Clinically Meaningful Comparisons Over Time: An Approach to Measuring Patient Similarity based on Subsequence Alignment
• Sharing pattern submodels for prediction with missing values
• Quality control, data cleaning, imputation
• Synthetic Epileptic Brain Activities Using Generative Adversarial Networks
• Real-Time Workload Classification during Driving using HyperNetworks
• Deep Mixed Effect Model using Gaussian Processes: A Personalized and Reliable Prediction for Healthcare
• Long-term Blood Pressure Prediction with Deep Recurrent Neural Networks
• Improving Irregularly Sampled Time Series Learning with Dense Descriptors of Time
• DynImp: Dynamic Imputation for Wearable Sensing Data Through Sensory and Temporal Relatedness
• Bayesian Recurrent Framework for Missing Data Imputation and Prediction with Clinical Time Series
• CoSTI: Consistency Models for (a faster) Spatio-Temporal Imputation
• A Survey of Challenges and Opportunities in Sensing and Analytics for Cardiovascular Disorders
• Toward Reducing Crop Spoilage and Increasing Small Farmer Profits in India: a Simultaneous Hardware and Software Solution
• Neural Network-based Object Classification by Known and Unknown Features (Based on Text Queries)
• Dynamic Gaussian Mixture based Deep Generative Model For Robust Forecasting on Sparse Multivariate Time Series
• Medical data wrangling with sequential variational autoencoders