Modeling Clinical Time Series Using Gaussian Process Sequences

Hauskrecht, Miloš; Liu, Zitao; Wu, Lei

doi:10.1137/1.9781611972832.69

Cited by 15 publications

(12 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The covariance matrix has the ability to incorporate a kernel or function to modify the functionality, often smoothing or bring periodicity to the behavior [21]. The correct covariance function can increase when it is in regions which are further away from previous regions of known values, and thus shrinks when near [22]. The constant basis will be used for the function in this analysis.…”

Section: Gaussian Process Regressionmentioning

confidence: 99%

Representation of Clinical Information in Outpatient Oncology for Prognosis Using Regression

Winikus¹,

Brown²

2016

Adv. sci. technol. eng. syst. j.

View full text Add to dashboard Cite

The determination of length of survival, or prognosis, is often viewed through statistical hazard models or with respect to a future reference time point in a classification approach (e.g., survival after 2 or 5 years). In this research, regression was used to determine a patient's prognosis. Also, multiple behavioral representations of clinical data, including difference trends and splines, are considered for predictor variables, which is different from demographic and tumor characteristics often used. With this approach the amount of clinical samples considered from the available patient data in the model in conjunction with the behavioral representation was explored. The models with the best prognostic performance had data representations that included limited clinical samples and some behavioral interpretations.

show abstract

Section: Gaussian Process Regressionmentioning

confidence: 99%

Representation of Clinical Information in Outpatient Oncology for Prognosis Using Regression

Winikus¹,

Brown²

2016

Adv. sci. technol. eng. syst. j.

View full text Add to dashboard Cite

show abstract

“…[11] The time-stamped EHR observations, in particular, are often acquired asynchronously (i.e., measured at different time instants and sampled irregularly in time), sparse, and include heterogeneous longitudinal data (often both time points and time intervals), thus provide fundamental challenges for directly applying common temporal analysis methods. [11][12][13][14][15][16] For example, the record dates associated with diagnoses codes in EHRs often reflect when the diagnoses were made by clinicians, not the actual onset of the disease.…”

Section: Introductionmentioning

confidence: 99%

High-throughput Phenotyping with Temporal Sequences

Estiri

McCoy

Wagholikar

et al. 2019

Preprint

View full text Add to dashboard Cite

Electronic health records (EHRs) contain important temporal information about progression of disease and treatment outcomes. The objective of this paper is to propose and test a high-throughput approach for phenotyping using temporal sequences of EHR observations to derive predictive and interpretable data representations. Using expert labeled clinical data from a congestive heart failure (CHF) cohort, we applied a four-phase framework to mine new data representations from temporal sequences of EHR observations, reduce dimensionality of the feature space, evaluate feature importance, and classify CHF patients with CHF. We found that sequenced features significantly outperform raw EHR features in classifying heart failure patients. Compared to the gold standard created by a panel of domain experts, the best classifier using sequences of diagnosis and medication observations classified heart failure status with a receiver operating characteristic area under the curve of 0.926. Compared with their raw counterparts, the top sequenced features have lower prevalence in medical records, but present more useful information for classification as the majority of those who have the sequences are likely to have the disease. Raw EHR data are not direct reflections of true health states as they also reflect various healthcare processes. Our results demonstrated that data representations obtained from sequencing EHR observations can present novel insight about the progression of disease that is difficult to discern when clinical observation histories are treated independently. Harnessing knowledge of disease progression through temporal sequencing of EHR observations can improve computational disease classification or phenotyping.

show abstract

“…We refer to our approach as GPPM because it is based on the flexible Bayesian nonparametric multivariate regression method Gaussian process regression (GPR; Rasmussen & Williams, 2006). GPR is a popular statistical learning method that has been applied successfully for the analysis of time series data from fields such as astronomy (Damouras, 2008;Roberts et al, 2013), meteorology (Roberts et al, 2013), economics (Damouras, 2008;Roberts et al, 2013), biology (Saatçi, Turner, & Rasmussen, 2010), medicine (Brahim-Belhouari & Bermak, 2004;Liu, Wu, & Hauskrecht, 2013), and neuroimaging (Ziegler, Ridgway, Dahnke, & Gaser, 2014). We refer to GPR as applied in the analysis of time series as temporal GPR.…”

Section: Introductionmentioning

confidence: 99%

Gaussian Process Panel Modeling – Kernel-Based Longitudinal Modeling

Karch

Brandmaier

Voelkle

2017

Preprint

View full text Add to dashboard Cite

Longitudinal panel data obtained from multiple individuals measured at multiple time points are crucial for psychological research. To analyze such data, a variety of modeling approaches such as hierarchical linear modeling or linear structural equation modeling are available. Such traditional parametric approaches are based on a relatively strong set of assumptions, which are often not met in practice. We present a flexible modeling approach for longitudinal data that is based on the Bayesian statistical learning method Gaussian Process Regression. We term this novel approach Gaussian Process Panel Modeling (GPPM). We show that GPPM subsumes most common modeling approaches for longitudinal data such as linear structural equation models and state-space models as special cases but also extends the space of expressible models beyond them. GPPM offers great flexibility in model specification, facilitates both parametric and nonparametric modeling in a single framework, enables continuous-time modeling as well as person-specific predictions, and offers a modular system that allows the user to piece together hypotheses about change by selecting from and combining predefined types of trajectories or dynamics. We demonstrate the utility of GPPM based on a selection of models and data sets.

show abstract

Modeling Clinical Time Series Using Gaussian Process Sequences

Cited by 15 publications

References 17 publications

Representation of Clinical Information in Outpatient Oncology for Prognosis Using Regression

Representation of Clinical Information in Outpatient Oncology for Prognosis Using Regression

High-throughput Phenotyping with Temporal Sequences

Gaussian Process Panel Modeling – Kernel-Based Longitudinal Modeling

Contact Info

Product

Resources

About