Artificial neural networks accurately predict mortality in patients with nonvariceal upper GI bleeding.

BACKGROUND Risk stratification systems that accurately identify patients with a high risk for bleeding through the use of clinical predictors of mortality before endoscopic examination are needed. Computerized (artificial) neural networks (ANNs) are adaptive tools that may improve prognostication. OBJECTIVE To assess the capability of an ANN to predict mortality in patients with nonvariceal upper GI bleeding and compare the predictive performance of the ANN with that of the Rockall score. DESIGN Prospective, multicenter study. SETTING Academic and community hospitals. PATIENTS This study involved 2380 patients with nonvariceal upper GI bleeding. INTERVENTION Upper GI endoscopy. MAIN OUTCOME MEASUREMENTS The primary outcome variable was 30-day mortality, defined as any death occurring within 30 days of the index bleeding episode. Other outcome variables were recurrent bleeding and need for surgery. RESULTS We performed analysis of certified outcomes of 2380 patients with nonvariceal upper GI bleeding. The Rockall score was compared with a supervised ANN (TWIST system, Semeion), adopting the same result validation protocol with random allocation of the sample in training and testing subsets and subsequent crossover. Overall, death occurred in 112 cases (4.70%). Of 68 pre-endoscopic input variables, 17 were selected and used by the ANN versus 16 included in the Rockall score. The sensitivity of the ANN-based model was 83.8% (76.7-90.8) versus 71.4% (62.8-80.0) for the Rockall score. Specificity was 97.5 (96.8-98.2) and 52.0 (49.8 4.2), respectively. Accuracy was 96.8% (96.0-97.5) versus 52.9% (50.8-55.0) (P<.001). The predictive performance of the ANN-based model for prediction of mortality was significantly superior to that of the complete Rockall score (area under the curve 0.95 [0.92-0.98] vs 0.67 [0.65-0.69]; P<.001). LIMITATIONS External validation on a subsequent independent population is needed, patients with variceal bleeding and obscure GI hemorrhage are excluded. CONCLUSION In patients with nonvariceal upper GI bleeding, ANNs are significantly superior to the Rockall score in predicting the risk of death.

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

[2]  R. Cestari,et al.  Predicting Mortality in Non-Variceal Upper Gastrointestinal Bleeders: Validation of the Italian PNED Score and Prospective Comparison With the Rockall Score , 2010, The American Journal of Gastroenterology.

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

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

[5]  J. Tijssen,et al.  Acute upper GI bleeding: did anything change? Time trend analysis of incidence and outcome of acute upper GI bleeding between 1993/1994 and 2000. , 2003, The American journal of gastroenterology.

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

[7]  R. Cestari,et al.  Predictive Factors of Mortality From Nonvariceal Upper Gastrointestinal Hemorrhage: A Multicenter Study , 2008, The American Journal of Gastroenterology.

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

[9]  R. Hunt,et al.  International Consensus Recommendations on the Management of Patients With Nonvariceal Upper Gastrointestinal Bleeding , 2010, Annals of Internal Medicine.

[10]  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.

[11]  L. Fabris,et al.  Changing trends in acute upper-GI bleeding: a population-based study. , 2009, Gastrointestinal endoscopy.

[12]  A. Barkun,et al.  Management of acute bleeding from a peptic ulcer. , 2008, The New England journal of medicine.

[13]  J. Scolapio.  Causes of Mortality in Patients With Peptic Ulcer Bleeding: A Prospective Cohort Study of 10,428 Cases , 2010 .

[14]  P. Chiu,et al.  Predicting poor outcome from acute upper gastrointestinal hemorrhage. , 2009, Gastroenterology Clinics of North America.

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

[16]  V. Wong,et al.  Predicting mortality in patients with bleeding peptic ulcers after therapeutic endoscopy. , 2009, Clinical gastroenterology and hepatology : the official clinical practice journal of the American Gastroenterological Association.

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

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

[19]  David W. McFadden,et al.  Articles in Press , 2008 .

[20]  Paolo Massimo Buscema,et al.  Genetic doping algorithm (GenD): theory and applications , 2004, Expert Syst. J. Knowl. Eng..

[21]  Farees T. Farooq,et al.  Artificial neural network as a predictive instrument in patients with acute nonvariceal upper gastrointestinal hemorrhage. , 2008, Gastroenterology.

[22]  James A Smith,et al.  Selective outpatient management of upper gastrointestinal bleeding in the elderly , 1999, American Journal of Gastroenterology.

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

[24]  G. Tytgat,et al.  Acute upper gastrointestinal bleeding in the Amsterdam area: incidence, diagnosis, and clinical outcome. , 1997, The American journal of gastroenterology.

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

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