Artificial neural network as a predictive instrument in patients with acute nonvariceal upper gastrointestinal hemorrhage.

BACKGROUND & AIMS Triage of patients with acute upper gastrointestinal hemorrhage (UGIH) has traditionally required urgent upper endoscopy. The aim of this study is to evaluate the use of artificial neural network for nonendoscopic triage. METHODS A cohort of 387 patients was used to train (n = 194) and internally validate (n = 193) the neural network, which was then externally validated in 200 patients and compared with the clinical and complete Rockall score. Two outcome variables were assessed: major stigmata of recent hemorrhage and need for endoscopic therapy. Patient cohort data from 2 independent tertiary-care medical centers were prospectively collected. Adult patients hospitalized at both sites during the same time period with a primary diagnosis of acute nonvariceal UGIH. RESULTS In predicting the 2 measured outcomes, sensitivity of neural network was >80%, with high negative predictive values (92-96%) in both cohorts but with lower specificity in the external cohort. Both Rockall scores had adequate sensitivity (>80%) but poor specificity (<40%) at outcome prediction. Comparing areas under receiver operating characteristic curves, the clinical Rockall score was significantly inferior to neural network in both cohorts (</=0.65 vs. >/= 0.78), while in the external cohort, neural network performed similarly to the complete Rockall score (>/= 0.78). CONCLUSIONS In acute nonvariceal UGIH, artificial neural network (nonendoscopic triage) performed as well as the complete Rockall score (endoscopic triage) at predicting stigmata of recent hemorrhage and need for endoscopic therapy, even when tested in an external patient population.

[1]  A. Barkun,et al.  Validation of the Rockall scoring system for outcomes from non-variceal upper gastrointestinal bleeding in a Canadian setting. , 2006, World journal of gastroenterology.

[2]  L. Targownik,et al.  Trends in management and outcomes of acute nonvariceal upper gastrointestinal bleeding: 1993-2003. , 2006, Clinical gastroenterology and hepatology : the official clinical practice journal of the American Gastroenterological Association.

[3]  T. Tham,et al.  Predicting outcome of acute non-variceal upper gastrointestinal haemorrhage without endoscopy using the clinical Rockall Score , 2006, Postgraduate Medical Journal.

[4]  I. Gralnek,et al.  Nonvariceal upper gastrointestinal bleeding: epidemiology and diagnosis. , 2005, Gastroenterology clinics of North America.

[5]  Xue Bai,et al.  A study of sample size with neural network , 2004, Proceedings of 2004 International Conference on Machine Learning and Cybernetics (IEEE Cat. No.04EX826).

[6]  Ananya Das,et al.  Prediction of outcome of acute GI hemorrhage: a review of risk scores and predictive models. , 2004, Gastrointestinal endoscopy.

[7]  I. Gralnek,et al.  Incremental value of upper endoscopy for triage of patients with acute non-variceal upper-GI hemorrhage. , 2004, Gastrointestinal endoscopy.

[8]  Alan Barkun,et al.  Consensus Recommendations for Managing Patients with Nonvariceal Upper Gastrointestinal Bleeding , 2003, Annals of Internal Medicine.

[9]  Amitabh Chak,et al.  Prediction of outcome in acute lower-gastrointestinal haemorrhage based on an artificial neural network: internal and external validation of a predictive model , 2003, The Lancet.

[10]  J. Reitsma,et al.  Acute upper GI bleeding: did anything change? , 2003, American Journal of Gastroenterology.

[11]  O. Blatchford,et al.  A risk score to predict need for treatment for uppergastrointestinal haemorrhage , 2000, The Lancet.

[12]  C. Prakash,et al.  Acute gastrointestinal bleeding. , 2000, The Medical clinics of North America.

[13]  Alan Morris.  Developing and Implementing Computerized Protocols for Standardization of Clinical Decisions , 2000, Annals of Internal Medicine.

[14]  P. Romano,et al.  Endoscopy-based triage significantly reduces hospitalization rates and costs of treating upper GI bleeding: a randomized controlled trial. , 1999, Gastrointestinal endoscopy.

[15]  R. F. Wagner,et al.  Classifier design for computer-aided diagnosis: effects of finite sample size on the mean performance of classical and neural network classifiers. , 1999, Medical physics.

[16]  C. Terwee,et al.  Validation of the Rockall risk scoring system in upper gastrointestinal bleeding , 1999, Gut.

[17]  R. Haynes,et al.  Effects of computer-based clinical decision support systems on physician performance and patient outcomes: a systematic review. , 1998, JAMA.

[18]  D. Classen,et al.  Clinical decision support systems to improve clinical practice and quality of care. , 1998, JAMA.

[19]  Simon Haykin,et al.  Neural Networks: A Comprehensive Foundation , 1998 .

[20]  T. C. Northfield,et al.  Selection of patients for early discharge or outpatient care after acute upper gastrointestinal haemorrhage , 1996, The Lancet.

[21]  N. Shaheen,et al.  The baylor score does not predict rebleeding following endoscopic therapy for acute peptic ulcer hemorrhage , 1996 .

[22]  H B Devlin,et al.  Risk assessment after acute upper gastrointestinal haemorrhage. , 1996, Gut.

[23]  W. Baxt.  Application of artificial neural networks to clinical medicine , 1995, The Lancet.

[24]  Patrick van der Smagt,et al.  Introduction to neural networks , 1995, The Lancet.

[25]  T. Rockall,et al.  Variation in outcome after acute upper gastrointestinal haemorrhage , 1995, The Lancet.

[26]  D. Graham,et al.  Prospective validation of the Baylor bleeding score for predicting the likelihood of rebleeding after endoscopic hemostasis of peptic ulcers. , 1995, Gastrointestinal Endoscopy.

[27]  L. Laine,et al.  A prospective outcome study of patients with clot in an ulcer and the effect of irrigation. , 1995, Gastrointestinal endoscopy.

[28]  G. Longstreth.  Epidemiology of hospitalization for acute upper gastrointestinal hemorrhage: a population-based study. , 1995, The American journal of gastroenterology.

[29]  R. Haynes,et al.  Effects of Computer-based Clinical Decision Support Systems on Clinician Performance and Patient Outcome: A Critical Appraisal of Research , 1994, Annals of Internal Medicine.

[30]  W. Baxt.  Analysis of the clinical variables driving decision in an artificial neural network trained to identify the presence of myocardial infarction. , 1992, Annals of emergency medicine.

[31]  W. Baxt.  Use of an artificial neural network for the diagnosis of myocardial infarction. , 1991, Annals of internal medicine.

[32]  D. Gilbert,et al.  Epidemiology of upper gastrointestinal bleeding. , 1990, Gastrointestinal endoscopy.

[33]  E. DeLong,et al.  Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. , 1988, Biometrics.

[34]  J. Hanley,et al.  The meaning and use of the area under a receiver operating characteristic (ROC) curve. , 1982, Radiology.

[35]  R. Marmo,et al.  Outpatient management for low-risk nonvariceal upper GI bleeding: a randomized controlled trial. , 2002, Gastrointestinal endoscopy.

[36]  Christopher M. Bishop,et al.  Neural networks for pattern recognition , 1995 .

[37]  G. A. Miller.  THE PSYCHOLOGICAL REVIEW THE MAGICAL NUMBER SEVEN, PLUS OR MINUS TWO: SOME LIMITS ON OUR CAPACITY FOR PROCESSING INFORMATION 1 , 1956 .