Dataset Information

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

ABSTRACT: 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).

SUBMITTER: Ghassemi M

PROVIDER: S-EPMC4864016 | biostudies-literature | 2015 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

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

Ghassemi Marzyeh M Pimentel Marco A F MA Naumann Tristan T Brennan Thomas T Clifton David A DA Szolovits Peter P Feng Mengling M

Proceedings of the ... AAAI Conference on Artificial Intelligence. AAAI Conference on Artificial Intelligence 20150101

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 cli ...[more]

PMID: 27182460

Similar Datasets

Project description:BackgroundCurrent prognostic scores for status epilepticus (SE) may not be adequate for patients in ICU who usually have more severe systemic conditions or more refractory episodes of SE. We aimed to compare the prognostic performance of two SE scores, Status Epilepticus Severity Score (STESS) and Epidemiology-Based Mortality Score in Status Epilepticus (EMSE) score, with four systemic severity scores, Acute Physiology and Chronic Health Evaluation 2 (APACHE-2), Simplified Acute Physiology Score 2 (SAPS-2), Sequential Organ Failure Assessment (SOFA) score, and Inflammation, Nutrition, Consciousness, Neurologic function and Systemic condition (INCNS) score in critically ill patients with SE.MethodsThis retrospective observational study of a prospectively identified SE cohort was conducted in the ICU at a tertiary-care center. The area under the receiver operating characteristic curve (AUC), sensitivity, specificity, positive predictive value, negative predictive value, accuracy, and associations with outcomes of STESS, EMSE, INCNS, APACHE-2, SAPS-2, and SOFA score for the prediction of in-hospital mortality and no return to baseline condition were assessed.ResultsBetween January 2015 and December 2020, 166 patients with SE in ICU were included in the study. In predicting in-hospital death, APACHE-2 (0.72), SAPS-2 (0.73), and SOFA score (0.71) had higher AUCs than STESS (0.58) and EMSE (0.69). In predicting no return to baseline condition, the AUC of APACHE-2 (0.75) was the highest, and the AUC of INCNS (0.55) was the lowest. When the specificity approached 90%, the sensitivity values of these scores were not quite acceptable (< 40%). Neither SE scores nor systemic severity scores had desirable prognostic power. In the multivariate logistic regression analyses, the best combinations of scores always included at least one or more systemic severity scores.ConclusionsSTESS and EMSE were insufficient in outcome prediction for SE patients in ICU, and EMSE was marginally better than STESS. Systemic illness matters in ICU patients with SE, and SE scores should be modified to achieve better accuracy in this severely ill population. This study mostly refers to severely ill patients in the ICU.

Project description:OBJECTIVES:Delirium severity is independently associated with longer hospital stays, nursing home placement, and death in patients outside the ICU. Delirium severity in the ICU is not routinely measured because the available instruments are difficult to complete in critically ill patients. We designed our study to assess the reliability and validity of a new ICU delirium severity tool, the Confusion Assessment Method for the ICU-7 delirium severity scale. DESIGN:Observational cohort study. SETTING:Medical, surgical, and progressive ICUs of three academic hospitals. PATIENTS:Five hundred eighteen adult (? 18 yr) patients. INTERVENTIONS:None. MEASUREMENTS AND MAIN RESULTS:Patients received the Confusion Assessment Method for the ICU, Richmond Agitation-Sedation Scale, and Delirium Rating Scale-Revised-98 assessments. A 7-point scale (0-7) was derived from responses to the Confusion Assessment Method for the ICU and Richmond Agitation-Sedation Scale items. Confusion Assessment Method for the ICU-7 showed high internal consistency (Cronbach's ? = 0.85) and good correlation with Delirium Rating Scale-Revised-98 scores (correlation coefficient = 0.64). Known-groups validity was supported by the separation of mechanically ventilated and nonventilated assessments. Median Confusion Assessment Method for the ICU-7 scores demonstrated good predictive validity with higher odds (odds ratio = 1.47; 95% CI = 1.30-1.66) of in-hospital mortality and lower odds (odds ratio = 0.8; 95% CI = 0.72-0.9) of being discharged home after adjusting for age, race, gender, severity of illness, and chronic comorbidities. Higher Confusion Assessment Method for the ICU-7 scores were also associated with increased length of ICU stay (p = 0.001). CONCLUSIONS:Our results suggest that Confusion Assessment Method for the ICU-7 is a valid and reliable delirium severity measure among ICU patients. Further research comparing it to other delirium severity measures, its use in delirium efficacy trials, and real-life implementation is needed to determine its role in research and clinical practice.

Project description:Human associated microbial communities exert tremendous influence over human health and disease. With modern metagenomic sequencing methods it is now possible to follow the relative abundance of microbes in a community over time. These microbial communities exhibit rich ecological dynamics and an important goal of microbial ecology is to infer the ecological interactions between species directly from sequence data. Any algorithm for inferring ecological interactions must overcome three major obstacles: 1) a correlation between the abundances of two species does not imply that those species are interacting, 2) the sum constraint on the relative abundances obtained from metagenomic studies makes it difficult to infer the parameters in timeseries models, and 3) errors due to experimental uncertainty, or mis-assignment of sequencing reads into operational taxonomic units, bias inferences of species interactions due to a statistical problem called "errors-in-variables". Here we introduce an approach, Learning Interactions from MIcrobial Time Series (LIMITS), that overcomes these obstacles. LIMITS uses sparse linear regression with boostrap aggregation to infer a discrete-time Lotka-Volterra model for microbial dynamics. We tested LIMITS on synthetic data and showed that it could reliably infer the topology of the inter-species ecological interactions. We then used LIMITS to characterize the species interactions in the gut microbiomes of two individuals and found that the interaction networks varied significantly between individuals. Furthermore, we found that the interaction networks of the two individuals are dominated by distinct "keystone species", Bacteroides fragilis and Bacteroided stercosis, that have a disproportionate influence on the structure of the gut microbiome even though they are only found in moderate abundance. Based on our results, we hypothesize that the abundances of certain keystone species may be responsible for individuality in the human gut microbiome.

Dataset Information

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

Publications

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

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets