Learning to Detect Sepsis with a Multitask Gaussian Process RNN Classifier

We present a scalable end-to-end classifier that uses streaming physiological and medication data to accurately predict the onset of sepsis, a life-threatening complication from infections that has high mortality and morbidity. Our proposed framework models the multivariate trajectories of continuous-valued physiological time series using multitask Gaussian processes, seamlessly accounting for the high uncertainty, frequent missingness, and irregular sampling rates typically associated with real clinical data. The Gaussian process is directly connected to a black-box classifier that predicts whether a patient will become septic, chosen in our case to be a recurrent neural network to account for the extreme variability in the length of patient encounters. We show how to scale the computations associated with the Gaussian process in a manner so that the entire system can be discriminatively trained end-to-end using backpropagation. In a large cohort of heterogeneous inpatient encounters at our university health system we find that it outperforms several baselines at predicting sepsis, and yields 19.4% and 55.5% improved areas under the Receiver Operating Characteristic and Precision Recall curves as compared to the NEWS score currently used by our hospital.

• Publication year

  2017

• Venue

  International Conference on Machine Learning

• Publication date

  2017-06-13

• Fields of study

  Medicine, Computer Science, Mathematics

• Identifiers
  arXiv 1706.04152
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Short Term

  2018cited by this paper
• Time to Treatment and Mortality during Mandated Emergency Care for Sepsis

  2017cited by this paper
• A Non-parametric Learning Method for Confidently Estimating Patient's Clinical State and Dynamics

  2016cited by this paper
• Using recurrent neural network models for early detection of heart failure onset

  2016cited by this paper
• Recurrent Neural Networks for Multivariate Time Series with Missing Values

  2016cited by this paper
• A scalable end-to-end Gaussian process adapter for irregularly sampled time series classification

  2016influential reference
• ForecastICU: A Prognostic Decision Support System for Timely Prediction of Intensive Care Unit Admission

  2016cited by this paper
• Scalable Joint Modeling of Longitudinal and Point Process Data for Disease Trajectory Prediction and Improving Management of Chronic Kidney Disease

  2016cited by this paper
• A Framework for Individualizing Predictions of Disease Trajectories by Exploiting Multi-Resolution Structure

  2015cited by this paper
• Multitask Gaussian Processes for Multivariate Physiological Time-Series Analysis

  2015cited by this paper
• A Multivariate Timeseries Modeling Approach to Severity of Illness Assessment and Forecasting in ICU with Sparse, Heterogeneous Clinical Data

  2015cited by this paper
• Efficient Learning of Continuous-Time Hidden Markov Models for Disease Progression

  2015cited by this paper
• Learning to Diagnose with LSTM Recurrent Neural Networks

  2015cited by this paper
• The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3)

  2015cited by this paper
• A targeted real-time early warning score (TREWScore) for septic shock

  2015cited by this paper
• Preconditioned Krylov Subspace Methods for Sampling Multivariate Gaussian Distributions

  2014cited by this paper
• Adam: A Method for Stochastic Optimization

  2014cited by this paper
• The ability of the National Early Warning Score (NEWS) to discriminate patients at risk of early cardiac arrest, unanticipated intensive care unit admission, and death.

  2013cited by this paper
• Auto-Encoding Variational Bayes

  2013cited by this paper
• Development and validation of a continuous measure of patient condition using the Electronic Medical Record

  2013cited by this paper
• Early classification on time series

  2011cited by this paper
• Lactate clearance vs central venous oxygen saturation as goals of early sepsis therapy: a randomized clinical trial.

  2010cited by this paper
• Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Critical Care Medicine. 1992.

  2009cited by this paper
• Multi-task Gaussian Process Prediction

  2007cited by this paper
• Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock*

  2006influential reference
• The value of Modified Early Warning Score (MEWS) in surgical in-patients: a prospective observational study.

  2006cited by this paper
• Multivariate geostatistics: an introduction with applications

  1998cited by this paper
• Long Short-Term Memory

  1997cited by this paper
• The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure

  1996cited by this paper
• Guidelines for the Use of Innovative Therapies in Sepsis

  1993cited by this paper
• Sepsis syndrome: a valid clinical entity. Methylprednisolone Severe Sepsis Study Group.

  1989cited by this paper
• Sepsis syndrome: a valid clinical entity

  1989cited by this paper

• Toward Automated Regulatory Decision-Making: AI-assisted Medical Device Risk Classification with Multimodal Transformers and Self-Training

  2026cites this paper
• Machine learning models explanations as interpretations of evidence: a theoretical framework of explainability and its implications on high-stakes biomedical decision-making

  2025cites this paper
• SepsisSuite: Beyond Risk Stratification - A Comparative Analysis of Deep Fusion vs. Expert Stacking for Prescriptive Sepsis AI

  2025cites this paper
• Comparison of Context-based Imputation for Missing Intracranial Pressure Signal

  2025cites this paper
• SurvBench: A Standardised Preprocessing Pipeline for Multi-Modal Electronic Health Record Survival Analysis

  2025cites this paper
• Optimal Look-back Horizon for Time Series Forecasting in Federated Learning

  2025cites this paper
• Data reuse enables cost-efficient randomized trials of medical AI models

  2025cites this paper
• Accurate prediction of sepsis from pediatric emergency department to PICU using a machine-learning model

  2025cites this paper
• ITNet: Irregular Timeseries Data Fusion with Attention Mechanisms

  2025cites this paper
• Generalized Signature Method for Multivariate Time Series: A Data-Driven Framework for Feature Extraction

  2025cites this paper
• Advancing Precision Oncology Through Modeling of Longitudinal and Multimodal Data

  2025cites this paper
• Predicting Septic Shock Using Explainable AI

  2025cites this paper
• SFC: a time series decomposition attention network with continuous nature for time series analysis

  2025cites this paper
• Towards Better Interpretability of Sepsis Prediction by Deep Neural Networks with Variable-Wise Attribution Maps

  2025cites this paper
• DynEdges-TGN: Dynamic-Edges-Based Temporal Graph Network for Early Sepsis Prediction

  2025cites this paper
• Dynamic Prediction and Intervention of Serum Sodium in Patients with Stroke Based on Attention Mechanism Model

  2025cites this paper
• Uncertainty-Inspired Multi-Task Learning in Arbitrary Scenarios of ECG Monitoring

  2025cites this paper
• Subtle variation in sepsis-III definitions markedly influences predictive performance within and across methods

  2024cites this paper
• DEEP AND MACHINE LEARNING MODELS FOR RECOGNIZING STATIC AND DYNAMIC GESTURES OF THE KAZAKH ALPHABET

  2024cites this paper
• MUSE-Net: Missingness-aware mUlti-branching Self-attention Encoder for Irregular Longitudinal Electronic Health Records

  2024cites this paper
• Predicting Solar Heat Production to Optimize Renewable Energy Usage

  2024cites this paper
• Augmenting Ground-Level PM2.5 Prediction via Kriging-Based Pseudo-Label Generation

  2024cites this paper
• DNA-T: Deformable Neighborhood Attention Transformer for Irregular Medical Time Series

  2024cites this paper
• Integration of Federated Learning and Blockchain in Healthcare: A Tutorial

  2024cites this paper
• Sequential Inference of Hospitalization ElectronicHealth Records Using Probabilistic Models

  2024cites this paper
• Dynamic Survival Analysis for Early Event Prediction

  2024cites this paper
• Selective Imputation for Multivariate Time Series Datasets With Missing Values

  2023cites this paper
• Clairvoyance: A Pipeline Toolkit for Medical Time Series

  2023cites this paper
• A Peephole Bidirectional the Long Short-Term Memory Network for Multi-Scale Imbalanced Time Series Classification

  2023cites this paper
• Risk Estimation for ICU Patients with Personalized Anomaly-Encoded Bedside Patient Data

  2023cites this paper
• Graph-Based Patient Representation for Multimodal Clinical Data: Addressing Data Heterogeneity

  2023cites this paper
• Self-attention with temporal prior: can we learn more from the arrow of time?

  2023cites this paper
• Time-Parameterized Convolutional Neural Networks for Irregularly Sampled Time Series

  2023cites this paper
• A New Acquisition Function for Multi-objective Bayesian Optimization: Correlated Probability of Improvement

  2023cites this paper
• Explaining a machine learning decision to physicians via counterfactuals

  2023cites this paper
• Factorial SDE for Multi-Output Gaussian Process Regression

  2023cites this paper
• Deployment of machine learning algorithms to predict sepsis: systematic review and application of the SALIENT clinical AI implementation framework

  2023cites this paper
• PoEMS: Policy Network-Based Early Warning Monitoring System for Sepsis in Intensive Care Units

  2023cites this paper
• NPRL: Nightly Profile Representation Learning for Early Sepsis Onset Prediction in ICU Trauma Patients

  2023cites this paper
• Daily Pain Prediction in Workplace Using Gaussian Processes

  2023cites this paper
• Data Streams: Investigating Data Structures for Multivariate Asynchronous Time Series Prediction Problems

  2023cites this paper
• Cross-center Early Sepsis Recognition by Medical Knowledge Guided Collaborative Learning for Data-scarce Hospitals

  2023cites this paper
• Learning to Decouple Complex Systems

  2023cites this paper
• Imputing Missing Observations with Time Sliced Synthetic Minority Oversampling Technique

  2022cites this paper
• Heteroscedastic Temporal Variational Autoencoder For Irregularly Sampled Time Series

  2022cites this paper
• Deep Kernel Learning for Mortality Prediction in the Face of Temporal Shift

  2022cites this paper
• Temporal Label Smoothing for Early Event Prediction

  2022cites this paper
• Predicting The Likelihood of Patients Developing Sepsis Based on Compound Ensemble Learning

  2022influential citation
• Manifestations of xenophobia in AI systems

  2022cites this paper
• ALRt: An Active Learning Framework for Irregularly Sampled Temporal Data

  2022cites this paper
• On the Importance of Clinical Notes in Multi-modal Learning for EHR Data

  2022cites this paper
• Prediction-Constrained Markov Models for Medical Time Series with Missing Data and Few Labels

  2022cites this paper
• A Framework for the Evaluation of Clinical Time Series Models

  2022cites this paper
• Forecasting Time Series in Healthcare With Gaussian Processes and Dynamic Time Warping Based Subset Selection

  2022cites this paper
• Self-explaining Hierarchical Model for Intraoperative Time Series

  2022cites this paper
• Prediction of Pediatric Sepsis Using a Deep Encoding Network with Cross Features

  2022cites this paper
• MACHINE LEARNING TO DEVELOP A MODEL THAT PREDICTS EARLY IMPENDING SEPSIS IN NEUROSURGICAL PATIENTS

  2022cites this paper
• mRisk

  2022cites this paper
• Temporal Label Smoothing for Early Prediction of Adverse Events

  2022cites this paper
• Who Goes First? Influences of Human-AI Workflow on Decision Making in Clinical Imaging

  2022cites this paper
• Machine Learning Models for Early Prediction of Sepsis on Large Healthcare Datasets

  2022cites this paper
• Multi-View Integrative Attention-Based Deep Representation Learning for Irregular Clinical Time-Series Data

  2022influential citation
• A multi-task Gaussian process self-attention neural network for real-time prediction of the need for mechanical ventilators in COVID-19 patients

  2022cites this paper
• Is In-hospital Meta-information Useful for Abstractive Discharge Summary Generation?

  2022cites this paper
• Unsupervised Probabilistic Models for Sequential Electronic Health Records

  2022cites this paper
• Optimizing Early Warning Classifiers to Control False Alarms via a Minimum Precision Constraint

  2022cites this paper
• Task-Synchronized Recurrent Neural Networks

  2022cites this paper
• ELDA: Learning Explicit Dual-Interactions for Healthcare Analytics

  2022cites this paper
• A Review of Recurrent Neural Network-Based Methods in Computational Physiology

  2022cites this paper
• The Secondary Use of Electronic Health Records for Data Mining: Data Characteristics and Challenges

  2022cites this paper
• Learning Optimal Summaries of Clinical Time-series with Concept Bottleneck Models

  2022cites this paper
• Improving Sepsis Prediction Model Generalization With Optimal Transport

  2022cites this paper
• Predicting sepsis in multi-site, multi-national intensive care cohorts using deep learning

  2021cites this paper
• Machine Learning for Biomedical Time Series Classification: From Shapelets to Deep Learning

  2021cites this paper
• Multi-view Integration Learning for Irregularly-sampled Clinical Time Series

  2021influential citation
• The Consequences of the Framing of Machine Learning Risk Prediction Models: Evaluation of Sepsis in General Wards

  2021cites this paper
• Real-time Prediction for Mechanical Ventilation in COVID-19 Patients using A Multi-task Gaussian Process Multi-objective Self-attention Network

  2021cites this paper
• Neural SDEs as Infinite-Dimensional GANs

  2021cites this paper
• A Temporal Kernel Approach for Deep Learning with Continuous-time Information

  2021cites this paper
• CARRNN: A Continuous Autoregressive Recurrent Neural Network for Deep Representation Learning From Sporadic Temporal Data

  2021cites this paper
• TOP-Net Prediction Model Using Bidirectional Long Short-term Memory and Medical-Grade Wearable Multisensor System for Tachycardia Onset: Algorithm Development Study

  2021cites this paper
• A review of irregular time series data handling with gated recurrent neural networks

  2021cites this paper
• Contextual explanation rules for neural clinical classifiers

  2021cites this paper
• Measuring the sensitivity of Gaussian processes to kernel choice

  2021cites this paper
• Semi-supervised Optimal Transport with Self-paced Ensemble for Cross-hospital Sepsis Early Detection

  2021influential citation
• Neural Natural Language Processing for Unstructured Data in Electronic Health Records: a Review

  2021cites this paper
• Generalization in Clinical Prediction Models: The Blessing and Curse of Measurement Indicator Variables

  2021cites this paper
• DBNet: a novel deep learning framework for mechanical ventilation prediction using electronic health records

  2021cites this paper
• Self-supervised Transformer for Multivariate Clinical Time-Series with Missing Values

  2021cites this paper
• Towards real-time diagnosis for pediatric sepsis using graph neural network and ensemble methods.

  2021cites this paper
• DAMA-Net: A Novel Predictive Model for Irregularly Asynchronously and Sparsely Sampled Multivariate Time Series

  2021cites this paper
• An explainable method for prediction of sepsis in ICUs using deep learning

  2021influential citation
• Hierarchical Gaussian Processes and Mixtures of Experts to Model COVID-19 Patient Trajectories

  2021cites this paper
• Improving Irregularly Sampled Time Series Learning with Time-Aware Dual-Attention Memory-Augmented Networks

  2021cites this paper
• A Generalised Signature Method for Multivariate Time Series Feature Extraction

  2021cites this paper
• The Framing of machine learning risk prediction models illustrated by evaluation of sepsis in general wards

  2021cites this paper
• Dynamic modeling of hospitalized COVID-19 patients reveals disease state–dependent risk factors

  2021cites this paper
• Assessing therapy response in patient-derived xenografts

  2021cites this paper
• An Interpretable Early Dynamic Sequential Predictor for Sepsis-Induced Coagulopathy Progression in the Real-World Using Machine Learning

  2021cites this paper
• Self-Supervised Transformer for Sparse and Irregularly Sampled Multivariate Clinical Time-Series

  2021cites this paper