Learning to Detect Sepsis with a Multitask Gaussian Process RNN Classifier

By Joseph D. Futoma, Sanjay Hariharan, Katherine A. Heller. 2017

194 citations recorded by the source corpus.

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.

Subfields: Artificial Intelligence, Epidemiology, Probabilistic & Bayesian Methods, 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

In Epidemiology

In Probabilistic & Bayesian Methods

In Medical & Health AI

In Signal Processing

Citations in the corpus

Neural Ordinary Differential Equations (2018, neutral): “Missing data can be addressed using generative time-series models (Álvarez and Lawrence, 2011; Futoma et al., 2017; Mei and Eisner, 2017; Soleimani et al., 2017a) or data imputation (Che et al., 2018).”
Neural Controlled Differential Equations for Irregular Time Series (2020, background): “For example [6, 7, 8, 9] incorporate exponential decay between observations, [10, 11] hybridise a Gaussian process with traditional neural network models, [12] approximate the underlying continuous-time process, and [13, 14] adapt recurrent neural networks by allowing some hidden state to evolve as…”
An Improved Multi-Output Gaussian Process RNN with Real-Time Validation for Early Sepsis Detection (2017, methodology): “Our proposed model improves upon our previous work in this area (Futoma et al. (2017)) that ties together a Multitask Gaussian Process with a Recurrent Neural Network classifier (MGP-RNN).”
GRU-ODE-Bayes: Continuous modeling of sporadically-observed time series (2019, methodology): “They have been used for smart imputation before a RNN or CNN architecture (Futoma et al., 2017; Moor et al., 2019), for modelling a hidden process in joint models (Soleimani et al., 2018), or to derive informative representations of time series (Ghassemi et al., 2015).”
Multi-Time Attention Networks for Irregularly Sampled Time Series (2020, background): “To respond to these challenges, there has been significant progress over the last decade on building and adapting machine learning models that can better capture the structure of irregularly sampled multivariate time series (Marlin et al., 2012; Li & Marlin, 2015; 2016; Lipton et al., 2016; Futoma e”
Latent ODEs for Irregularly-Sampled Time Series (2019, methodology): “Others have used a separate interpolation network [Shukla and Marlin, 2019], Gaussian processes [Futoma et al., 2017], or generative adversarial networks [Luo et al., 2018] to perform interpolation and imputation prior to running an RNN on time-discretized inputs.”
Predicting in-hospital mortality and unanticipated admissions to the intensive care unit using routinely collected blood tests and vital signs: Development and validation of a multivariable model (2018, background): “However, while others have used more complex machine learning approaches, such as neural networks [16,27], our study shows how a simple and easily interpretable score can still increase performance above that of NEWS alone.”
"The human body is a black box": supporting clinical decision-making with deep learning (2019, methodology): “The model generates risk scores every hour for every adult patient to detect sepsis [18, 19].”