Development of a model for prediction of survival in pediatric trauma patients: comparison of artificial neural networks and logistic regression.

BACKGROUND/PURPOSE There is a paucity of outcome prediction models for injured children. Using the National Pediatric Trauma Registry (NPTR), the authors developed an artificial neural network (ANN) to predict pediatric trauma death and compared it with logistic regression (LR). METHODS Patients in the NPTR from 1996 through 1999 were included. Models were generated using LR and ANN. A data search engine was used to generate the ANN with the best fit for the data. Input variables included anatomic and physiologic characteristics. There was a single output variable: probability of death. Assessment of the models was for both discrimination (ROC area under the curve) and calibration (Lemeshow-Hosmer C-Statistic). RESULTS There were 35,385 patients. The average age was 8.1 +/- 5.1 years, and there were 1,047 deaths (3.0%). Both modeling systems gave excellent discrimination (ROC A(z): LR = 0.964, ANN = 0.961). However, LR had only fair calibration, whereas the ANN model had excellent calibration (L/H C stat: LR = 36, ANN = 10.5). CONCLUSIONS The authors were able to develop an ANN model for the prediction of pediatric trauma death, which yielded excellent discrimination and calibration exceeding that of logistic regression. This model can be used by trauma centers to benchmark their performance in treating the pediatric trauma population.

[1]  F. Harrell,et al.  Artificial neural networks improve the accuracy of cancer survival prediction , 1997, Cancer.

[2]  D Greenwood,et al.  An overview of neural networks. , 1991, Behavioral science.

[3]  H. C. Veldenz,et al.  Pediatric risk indicator: an objective measurement of childhood injury severity. , 1997, The Journal of trauma.

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

[5]  M. Pollack,et al.  Resource use, efficiency, and outcome prediction in pediatric intensive care of trauma patients. , 1990, The Journal of trauma.

[6]  M. Rotondo,et al.  A new approach to probability of survival scoring for trauma quality assurance. , 1992, The Journal of trauma.

[7]  M. Giger,et al.  Computer vision and artificial intelligence in mammography. , 1994, AJR. American journal of roentgenology.

[8]  C E Floyd,et al.  Artificial neural network for diagnosis of acute pulmonary embolism: effect of case and observer selection. , 1995, Radiology.

[9]  Robert H. Brook,et al.  Assessing the Performance of Mortality Prediction Models , 1993 .

[10]  L. Pitts,et al.  Outcome after severe head injury: an analysis of prediction based upon comparison of neural network versus logistic regression analysis. , 1997, Neurological research.

[11]  S. Lemeshow,et al.  As American as apple pie and APACHE. Acute Physiology and Chronic Health Evaluation. , 1998, Critical care medicine.

[12]  M. Mckenney,et al.  Trauma scoring systems: a review. , 1999, Journal of the American College of Surgeons.

[13]  J. Savino,et al.  An artificial neural network as a model for prediction of survival in trauma patients: validation for a regional trauma area. , 1998 .

[14]  M. D. Williams,et al.  Prediction of trauma mortality using a neural network. , 1997, The American surgeon.

[15]  J. Tepas,et al.  The pediatric trauma score as a predictor of injury severity in the injured child. , 1987, Journal of pediatric surgery.

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

[17]  Dan W. Patterson,et al.  Artificial Neural Networks: Theory and Applications , 1998 .

[18]  S Mitchell,et al.  Predicting Outcomes After Liver Transplantation A Connectionist Approach , 1994, Annals of surgery.

[19]  M. Ebell.  Artificial neural networks for predicting failure to survive following in-hospital cardiopulmonary resuscitation. , 1993, The Journal of family practice.

[20]  U E Ruttimann,et al.  The Pediatric Risk of Mortality III--Acute Physiology Score (PRISM III-APS): a method of assessing physiologic instability for pediatric intensive care unit patients. , 1997, The Journal of pediatrics.

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

[22]  Anthony Slater,et al.  Paediatric index of mortality (PIM): a mortality prediction model for children in intensive care , 1997, Intensive Care Medicine.

[23]  B. Gans,et al.  National Pediatric Trauma Registry. , 1989, Journal of Pediatric Surgery.

[24]  P. Gregory,et al.  The Pediatric Risk of Mortality (PRISM) Score and Injury Severity Score (ISS) for predicting resource utilization and outcome of intensive care in pediatric trauma. , 1999, Critical care medicine.

[25]  E. Hannan,et al.  Predicting inpatient mortality for pediatric trauma patients with blunt injuries: a better alternative. , 2000, Journal of pediatric surgery.

[26]  G. Clermont,et al.  Predicting hospital mortality for patients in the intensive care unit: A comparison of artificial neural networks with logistic regression models , 2001, Critical care medicine.

[27]  D. Hosmer,et al.  A review of goodness of fit statistics for use in the development of logistic regression models. , 1982, American journal of epidemiology.