An Improved Multi-Output Gaussian Process RNN with Real-Time Validation for Early Sepsis Detection

Sepsis is a poorly understood and potentially life-threatening complication that can occur as a result of infection. Early detection and treatment improves patient outcomes, and as such it poses an important challenge in medicine. In this work, we develop a flexible classifier that leverages streaming lab results, vitals, and medications to predict sepsis before it occurs. We model patient clinical time series with multi-output Gaussian processes, maintaining uncertainty about the physiological state of a patient while also imputing missing values. The mean function takes into account the effects of medications administered on the trajectories of the physiological variables. Latent function values from the Gaussian process are then fed into a deep recurrent neural network to classify patient encounters as septic or not, and the overall model is trained end-to-end using back-propagation. We train and validate our model on a large dataset of 18 months of heterogeneous inpatient stays from the Duke University Health System, and develop a new "real-time" validation scheme for simulating the performance of our model as it will actually be used. Our proposed method substantially outperforms clinical baselines, and improves on a previous related model for detecting sepsis. Our model's predictions will be displayed in a real-time analytics dashboard to be used by a sepsis rapid response team to help detect and improve treatment of sepsis.

[1]  Edwin V. Bonilla,et al.  Multi-task Gaussian Process Prediction , 2007, NIPS.

[2]  K. Wood,et al.  Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock* , 2006, Critical care medicine.

[3]  Mihaela van der Schaar,et al.  A Non-parametric Learning Method for Confidently Estimating Patient's Clinical State and Dynamics , 2016, NIPS.

[4]  David C. Kale,et al.  Modeling Missing Data in Clinical Time Series with RNNs , 2016 .

[5]  Jimeng Sun,et al.  RETAIN: An Interpretable Predictive Model for Healthcare using Reverse Time Attention Mechanism , 2016, NIPS.

[6]  Katherine A. Heller,et al.  Learning to Detect Sepsis with a Multitask Gaussian Process RNN Classifier , 2017, ICML.

[7]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[8]  J. Vincent,et al.  The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure , 1996, Intensive Care Medicine.

[9]  Alan E. Jones,et al.  Lactate clearance vs central venous oxygen saturation as goals of early sepsis therapy: a randomized clinical trial. , 2010, JAMA.

[10]  P. Pronovost,et al.  A targeted real-time early warning score (TREWScore) for septic shock , 2015, Science Translational Medicine.

[11]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[12]  W. Knaus,et al.  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, Chest.

[13]  Yoshua Bengio,et al.  A Recurrent Latent Variable Model for Sequential Data , 2015, NIPS.

[14]  S. Lemeshow,et al.  Time to Treatment and Mortality during Mandated Emergency Care for Sepsis , 2017, The New England journal of medicine.

[15]  Max Welling,et al.  Auto-Encoding Variational Bayes , 2013, ICLR.

[16]  Kai Li,et al.  Sparse Multi-Output Gaussian Processes for Medical Time Series Prediction , 2017 .

[17]  Jimeng Sun,et al.  Using recurrent neural network models for early detection of heart failure onset , 2016, J. Am. Medical Informatics Assoc..

[18]  T. Clemmer,et al.  Sepsis syndrome: a valid clinical entity , 1989 .

[19]  Peter Szolovits,et al.  A Multivariate Timeseries Modeling Approach to Severity of Illness Assessment and Forecasting in ICU with Sparse, Heterogeneous Clinical Data , 2015, AAAI.

[20]  Gary B. Smith,et al.  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. , 2013, Resuscitation.

[21]  David A. Clifton,et al.  Multitask Gaussian Processes for Multivariate Physiological Time-Series Analysis , 2015, IEEE Transactions on Biomedical Engineering.

[22]  Ricard Ferrer,et al.  Effectiveness of treatments for severe sepsis: a prospective, multicenter, observational study. , 2009, American journal of respiratory and critical care medicine.

[23]  M. Levy,et al.  2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference , 2003, Intensive care medicine.

[24]  Charles Elkan,et al.  Learning to Diagnose with LSTM Recurrent Neural Networks , 2015, ICLR.

[25]  Yan Liu,et al.  Recurrent Neural Networks for Multivariate Time Series with Missing Values , 2016, Scientific Reports.

[26]  T. Clemmer,et al.  Sepsis syndrome: a valid clinical entity. Methylprednisolone Severe Sepsis Study Group. , 1989, Critical care medicine.

[27]  R. Bellomo,et al.  The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). , 2016, JAMA.

[28]  Mihaela van der Schaar,et al.  ForecastICU: A Prognostic Decision Support System for Timely Prediction of Intensive Care Unit Admission , 2016, ICML.

[29]  Edwin V. Bonilla,et al.  Collaborative Multi-output Gaussian Processes , 2014, UAI.

[30]  J. Gardner-Thorpe,et al.  The value of Modified Early Warning Score (MEWS) in surgical in-patients: a prospective observational study. , 2006, Annals of the Royal College of Surgeons of England.

[31]  Benjamin M. Marlin,et al.  A scalable end-to-end Gaussian process adapter for irregularly sampled time series classification , 2016, NIPS.

[32]  Neil D. Lawrence,et al.  Kernels for Vector-Valued Functions: a Review , 2011, Found. Trends Mach. Learn..