Prediction of outcome in acute lower-gastrointestinal haemorrhage based on an artificial neural network: internal and external validation of a predictive model

Das, Ananya; Ben–Menachem, Tamir; Cooper, Gregory S.; Chak, Amitabh; Sivak, Michael; Gonet, Judith A.; Wong, Richard C.K.

doi:10.1016/s0140-6736(03)14568-0

Cited by 180 publications

(116 citation statements)

References 24 publications

Supporting

Mentioning

113

Contrasting

Order By: Relevance

“…ANNs have shown significant promise in modeling outcomes in different types of cancer [10][27], gastrointestinal bleeding [7], and multi-system trauma [8]. The use of ANNs to predict outcome of head trauma was limited to a single model, with a small sample size [9].…”

Section: Discussionmentioning

confidence: 99%

Informative sampling for large unbalanced data sets

Rughani

Tranmer

et al. 2008

Proceedings of the 10th Annual Conference Companion on Genetic and Evolutionary Computation

View full text Add to dashboard Cite

Selective sampling is a form of active learning which can reduce the cost of training by only drawing informative data points into the training set. This selected training set is expected to contain more information for modeling compared to random sampling, thus making modeling faster and more accurate. We introduce a novel approach to selective sampling, which is derived from the Estimation-Exploration Algorithm (EEA). The EEA is a coevolutionary algorithm that uses model disagreement to determine the significance of a training datum, and evolves a set of models only on the selected data. The algorithm in this paper trains a population of Artificial Neural Networks (ANN) on the training set, and uses their disagreement to seek new data for the training set. A medical data set called the National Trauma Data Bank (NTDB) is used to test the algorithm. Experiments show that the algorithm outperforms the equivalent algorithm using randomly-selected data and sampling evenly from each class. Finally, the selected training data reveals which features most affect outcome, allowing for both improved modeling and understanding of the processes that gave rise to the data.

show abstract

Section: Discussionmentioning

confidence: 99%

Informative sampling for large unbalanced data sets

Rughani

Tranmer

et al. 2008

Proceedings of the 10th Annual Conference Companion on Genetic and Evolutionary Computation

View full text Add to dashboard Cite

show abstract

“…These statistical methods have been described previously in the field of hepatic diseases [30][31][32]. The performance of these techniques was compared.…”

Section: Development Of Predictive Models Using Data Mining Analysismentioning

confidence: 99%

FibroGENE: A gene-based model for staging liver fibrosis

Eslam

Hashem

Romero–Gómez

et al. 2016

Journal of Hepatology

View full text Add to dashboard Cite

: A gene-based model for staging liver fibrosis, Journal of Hepatology (2015), doi: http://dx.doi.org/10. 1016/j.jhep.2015.11.008 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Background& aim: The extent of liver fibrosis predicts long term outcomes, and hence impacts

show abstract

“…Many other approaches are specific for acute upper or acute lower GIH but the source of the bleed is not always known prior to endoscopy. [19][20][21][22][23] One of the risk stratification systems for predicting the outcome of gastrointestinal (GI) bleeding has been introduced by Kollef et al, who identified five predictors of risk for in-hospital complications and suggested they could be used to triage patients with upper and patients with lower GI hemorrhage.…”

Section: Introductionmentioning

confidence: 99%

BLEED: a classification tool to predict outcomes in patients with upper and lower gastrointestinal hemorrhage

Huchchannavar¹,

Puttannavar²,

Hebsur³

2017

Int Surg J

View full text Add to dashboard Cite

Background: Acute gastrointestinal (GI) haemorrhage is a common clinical problem with diverse manifestations. Such bleeding may range from trivial to massive and can originate from virtually any region of the GI tract, including the pancreas, liver, and biliary tree. Several risk scoring systems have also been proposed to classify patients into high and low risk groups for complications, like re-bleeding or mortality, based on multivariate analyses. Kollef and colleagues identified the BLEED criteria: (a) ongoing Bleeding, (b) Low systolic blood pressure (BP), (c) Elevated prothrombin time (PT), (d) Erratic mental status, and (e) unstable comorbid Disease as risk factors for complication of GIH at any time during hospitalization after an initial 24 hours of stabilization. The objective of this study was to predict outcome according to a risk stratification BLEED criterion, independent of endoscopic findings. Methods: We studied all patients who presented with acute gastrointestinal bleeding to emergency department. patients with epistaxis, paranasal sinuses bleed, upper GI bleed secondary to endoscopic procedure, patients with chronic Anemia and those patients which admitted with Primary diagnosis other than UGIB were excluded. Patients meeting the BLEED criteria at their initial assessment were classified as high risk (66) and all others were categorized as low-risk (10). In-hospital complications were defined as recurrent UGIB, surgery to control the source of hemorrhage, hospital mortality, length of hospital stay and units of blood transfused. Results: There were 76 patients, with mean age of 46.37 years, 56 patients (73.3%) were case of Upper gastrointestinal bleeding, 20 patients (26.7%) were case of lower gastrointestinal bleeding. 66 (86.84%) of patients were categorized as high-risk patients and 10(13.2%) of patients were categorized as low risk patients. 14(21.1%) of patients were admitted in ICU ,13 Patients had undergone surgery (17.10%), 1 (1.5%) of patient had Re bleeding, nine (13%) had died. Stastical analysis showed significant association between components like low SBP (P=0.008), elevated prothrombin time (P=0.04), erratic mental status(P=0.001) and in hospital complications. All nine deaths were found in high risk group. Conclusions: BLEED criteria can be used as triage tool for stratifying the patients of acute gastrointestinal haemorrhage into high risk and low risk category without endoscopic findings and useful in predicting outcome in such patients and plan the treatment accordingly.

show abstract

Prediction of outcome in acute lower-gastrointestinal haemorrhage based on an artificial neural network: internal and external validation of a predictive model

Cited by 180 publications

References 24 publications

Informative sampling for large unbalanced data sets

Informative sampling for large unbalanced data sets

FibroGENE: A gene-based model for staging liver fibrosis

BLEED: a classification tool to predict outcomes in patients with upper and lower gastrointestinal hemorrhage

Contact Info

Product

Resources

About