Dataset Information

Real alerts and artifact classification in archived multi-signal vital sign monitoring data: implications for mining big data.

ABSTRACT: Huge hospital information system databases can be mined for knowledge discovery and decision support, but artifact in stored non-invasive vital sign (VS) high-frequency data streams limits its use. We used machine-learning (ML) algorithms trained on expert-labeled VS data streams to automatically classify VS alerts as real or artifact, thereby "cleaning" such data for future modeling. 634 admissions to a step-down unit had recorded continuous noninvasive VS monitoring data [heart rate (HR), respiratory rate (RR), peripheral arterial oxygen saturation (SpO2) at 1/20 Hz, and noninvasive oscillometric blood pressure (BP)]. Time data were across stability thresholds defined VS event epochs. Data were divided Block 1 as the ML training/cross-validation set and Block 2 the test set. Expert clinicians annotated Block 1 events as perceived real or artifact. After feature extraction, ML algorithms were trained to create and validate models automatically classifying events as real or artifact. The models were then tested on Block 2. Block 1 yielded 812 VS events, with 214 (26 %) judged by experts as artifact (RR 43 %, SpO2 40 %, BP 15 %, HR 2 %). ML algorithms applied to the Block 1 training/cross-validation set (tenfold cross-validation) gave area under the curve (AUC) scores of 0.97 RR, 0.91 BP and 0.76 SpO2. Performance when applied to Block 2 test data was AUC 0.94 RR, 0.84 BP and 0.72 SpO2. ML-defined algorithms applied to archived multi-signal continuous VS monitoring data allowed accurate automated classification of VS alerts as real or artifact, and could support data mining for future model building.

SUBMITTER: Hravnak M

PROVIDER: S-EPMC4821824 | biostudies-literature | 2016 Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Real alerts and artifact classification in archived multi-signal vital sign monitoring data: implications for mining big data.

Hravnak Marilyn M Chen Lujie L Dubrawski Artur A Bose Eliezer E Clermont Gilles G Pinsky Michael R MR

Journal of clinical monitoring and computing 20151005 6

Huge hospital information system databases can be mined for knowledge discovery and decision support, but artifact in stored non-invasive vital sign (VS) high-frequency data streams limits its use. We used machine-learning (ML) algorithms trained on expert-labeled VS data streams to automatically classify VS alerts as real or artifact, thereby "cleaning" such data for future modeling. 634 admissions to a step-down unit had recorded continuous noninvasive VS monitoring data [heart rate (HR), resp ...[more]

PMID: 26438655

Similar Datasets

Project description:Viruses of microbes impact all ecosystems where microbes drive key energy and substrate transformations including the oceans, humans and industrial fermenters. However, despite this recognized importance, our understanding of viral diversity and impacts remains limited by too few model systems and reference genomes. One way to fill these gaps in our knowledge of viral diversity is through the detection of viral signal in microbial genomic data. While multiple approaches have been developed and applied for the detection of prophages (viral genomes integrated in a microbial genome), new types of microbial genomic data are emerging that are more fragmented and larger scale, such as Single-cell Amplified Genomes (SAGs) of uncultivated organisms or genomic fragments assembled from metagenomic sequencing. Here, we present VirSorter, a tool designed to detect viral signal in these different types of microbial sequence data in both a reference-dependent and reference-independent manner, leveraging probabilistic models and extensive virome data to maximize detection of novel viruses. Performance testing shows that VirSorter's prophage prediction capability compares to that of available prophage predictors for complete genomes, but is superior in predicting viral sequences outside of a host genome (i.e., from extrachromosomal prophages, lytic infections, or partially assembled prophages). Furthermore, VirSorter outperforms existing tools for fragmented genomic and metagenomic datasets, and can identify viral signal in assembled sequence (contigs) as short as 3kb, while providing near-perfect identification (>95% Recall and 100% Precision) on contigs of at least 10kb. Because VirSorter scales to large datasets, it can also be used in "reverse" to more confidently identify viral sequence in viral metagenomes by sorting away cellular DNA whether derived from gene transfer agents, generalized transduction or contamination. Finally, VirSorter is made available through the iPlant Cyberinfrastructure that provides a web-based user interface interconnected with the required computing resources. VirSorter thus complements existing prophage prediction softwares to better leverage fragmented, SAG and metagenomic datasets in a way that will scale to modern sequencing. Given these features, VirSorter should enable the discovery of new viruses in microbial datasets, and further our understanding of uncultivated viral communities across diverse ecosystems.

Dataset Information

Real alerts and artifact classification in archived multi-signal vital sign monitoring data: implications for mining big data.

Publications

Real alerts and artifact classification in archived multi-signal vital sign monitoring data: implications for mining big data.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets