A Multivariate Timeseries Modeling Approach to Severity of Illness Assessment and Forecasting in ICU with Sparse, Heterogeneous Clinical Data

By Marzyeh Ghassemi, Marco A. F. Pimentel, Tristan Naumann, Thomas Brennan, David A. Clifton, Peter Szolovits, Mengling Feng. 2015

238 citations recorded by the source corpus.

The ability to determine patient acuity (or severity of illness) has immediate practical use for clinicians. We evaluate the use of multivariate timeseries modeling with the multi-task Gaussian process (GP) models using noisy, incomplete, sparse, heterogeneous and unevenly-sampled clinical data, including both physiological signals and clinical notes. The learned multi-task GP (MTGP) hyperparameters are then used to assess and forecast patient acuity. Experiments were conducted with two real clinical data sets acquired from ICU patients: firstly, estimating cerebrovascular pressure reactivity, an important indicator of secondary damage for traumatic brain injury patients, by learning the interactions between intracranial pressure and mean arterial blood pressure signals, and secondly, mortality prediction using clinical progress notes. In both cases, MTGPs provided improved results: an MTGP model provided better results than single-task GP models for signal interpolation and forecasting (0.91 vs 0.69 RMSE), and the use of MTGP hyperparameters obtained improved results when used as additional classification features (0.812 vs 0.788 AUC).

Subfields: Biomedical Engineering, Epidemiology, Surgery

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biomedical Engineering

In Epidemiology

In Surgery

Citations in the corpus

Natural Language Processing for EHR-Based Computational Phenotyping (2018, background): “[106], [142], [178], [179], applied topic models and extended tensor-based topic models to learn better coherent features.”
A Reinforcement Learning Approach to Weaning of Mechanical Ventilation in Intensive Care Units (2017, methodology): “One commonly used approach to resolve missing and irregularly sampled time series data is Gaussian processes (GPs, [Stegle et al., 2008, Dürichen et al., 2015, Ghassemi et al., 2015]).”
Learning to Diagnose with LSTM Recurrent Neural Networks (2015, background): “Gaussian processes are popular because they can directly handle irregular sampling and encode prior knowledge via choice of covariance functions between time steps and across variables (Marlin et al., 2012; Ghassemi et al., 2015).”
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).”
Learning Treatment-Response Models from Multivariate Longitudinal Data (2017, background): “For example, Ghassemi et al. (2015) use multi-task GPs, Schulam and Saria (2016) use coupled GPs, and Liu and Hauskrecht (2016) combine state-space models with GPs.”
Scalable Joint Models for Reliable Uncertainty-Aware Event Prediction (2017, methodology): “[19] use multi-task Gaussian processes to model multiple longitudinal series and use features estimated from the resulting fitted data to predict occurrence of an event.”
An Optimal Policy for Patient Laboratory Tests in Intensive Care Units (2018, unknown): “h is two-fold: rst, we build an interpretable model to forecast future patient states based on past observations, including uncertainty quantication. We adapt multi-output Gaussian processes (MOGPs; [9, 10]) to learn the patient state transition dynamics from a patient cohort including sparse and ir”
Sparse Multi-Output Gaussian Processes for Medical Time Series Prediction (2017, methodology): “In contrast to related work (Ghassemi et al., 2015b; Dürichen et al., 2015), our kernel uses a mixture of flexible spectral kernels (Wilson and Adams, 2013), allowing periodic behavior and both short-term and long-term dependencies within and across the clinical covariates over time.”
Benchmark of Deep Learning Models on Large Healthcare MIMIC Datasets (2017, background): “The availability of large healthcare databases such as Medical Information Mart for Intensive Care (MIMIC-II and III) databases [4, 5] has accelerated the research in this important area as evidenced by a lot of recent publications [6, 7, 8, 2, 9, 10, 11, 1, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 2”
Dynamic Measurement Scheduling for Adverse Event Forecasting using Deep RL (2018, background): “t experience. Recently developed time-series forecasting models solve the much needed problem of predicting early detection of adverse events (e.g. sepsis) based on sparse and irregular measurements (Ghassemi et al., 2015; Soleimani et al., 2017; Futoma et al., 2017). However, the timing of these me”
Learning to Detect Sepsis with a Multitask Gaussian Process RNN Classifier (2017, methodology): “For instance (Ghassemi et al., 2015) and (Durichen et al., 2015) used a similar model to ours, but instead focused more on forecasting of vitals to predict clinical instability, whereas our task is a binary classiﬁcation to identify sepsis early.”
An Improved Multi-Output Gaussian Process RNN with Real-Time Validation for Early Sepsis Detection (2017, background): “For instance (Ghassemi et al. (2015)) and (Durichen et al. (2015)) use multitask Gaussian processes and (Cheng et al.) use more complex multi-output Gaussian processes, with the focus in all on forecasting future vitals rather than predicting an event.”
Machine Learning and Decision Support in Critical Care (2016, methodology): “[200] used a multistep pipeline to predict ICU mortality.”
Multitask learning and benchmarking with clinical time series data (2017, background): “estive heart failure; nonhypertensive (9) Coronary atherosclerosis and related (10) Diabetes mellitus with complications (11) Diabetes mellitus without complication (12) Disorders of lipid metabolism (13) Essential hypertension (14) Fluid and electrolyte disorders (15) Gastrointestinal hemorrhage (1”