Advantages and disadvantages of using artificial neural networks versus logistic regression for predicting medical outcomes.

Artificial neural networks are algorithms that can be used to perform nonlinear statistical modeling and provide a new alternative to logistic regression, the most commonly used method for developing predictive models for dichotomous outcomes in medicine. Neural networks offer a number of advantages, including requiring less formal statistical training, ability to implicitly detect complex nonlinear relationships between dependent and independent variables, ability to detect all possible interactions between predictor variables, and the availability of multiple training algorithms. Disadvantages include its "black box" nature, greater computational burden, proneness to overfitting, and the empirical nature of model development. An overview of the features of neural networks and logistic regression is presented, and the advantages and disadvantages of using this modeling technique are discussed.

[1]  K A Spackman.  Combining logistic regression and neural networks to create predictive models. , 1992, Proceedings. Symposium on Computer Applications in Medical Care.

[2]  K. Liestøl,et al.  Survival analysis and neural nets. , 1994, Statistics in medicine.

[3]  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.

[4]  Geoffrey E. Hinton,et al.  How neural networks learn from experience. , 1992, Scientific American.

[5]  Halbert White,et al.  Learning in Artificial Neural Networks: A Statistical Perspective , 1989, Neural Computation.

[6]  M L Astion,et al.  Application of neural networks to the classification of giant cell arteritis. , 1994, Arthritis and rheumatism.

[7]  D J Spiegelhalter,et al.  Probabilistic prediction in patient management and clinical trials. , 1986, Statistics in medicine.

[8]  M. Rubenfire,et al.  Neural network in the clinical diagnosis of acute pulmonary embolism. , 1993, Chest.

[9]  Adam Blum,et al.  Neural Networks in C++: An Object-Oriented Framework for Building Connectionist Systems , 1992 .

[10]  J. Tu,et al.  Use of a neural network as a predictive instrument for length of stay in the intensive care unit following cardiac surgery. , 1993, Computers and biomedical research, an international journal.

[11]  Geoffrey E. Hinton,et al.  Learning representations by back-propagating errors , 1986, Nature.

[12]  D. Hosmer,et al.  Applied Logistic Regression , 1991 .

[13]  Bruce W. Colletti,et al.  Artificial intelligence versus logistic regression statistical modelling to predict cardiac complications after noncardiac surgery , 1994, Clinical Cardiology.

[14]  T. Buchman,et al.  A comparison of statistical and connectionist models for the prediction of chronicity in a surgical intensive care unit , 1994, Critical care medicine.

[15]  K A Spackman.  Maximum likelihood training of connectionist models: comparison with least squares back-propagation and logistic regression. , 1991, Proceedings. Symposium on Computer Applications in Medical Care.

[16]  H. Keselman,et al.  Backward, forward and stepwise automated subset selection algorithms: Frequency of obtaining authentic and noise variables , 1992 .

[17]  Richard Lippmann,et al.  Neural Network Classifiers Estimate Bayesian a posteriori Probabilities , 1991, Neural Computation.

[18]  J. Shao.  Linear Model Selection by Cross-validation , 1993 .

[19]  H. Sox,et al.  Clinical prediction rules. Applications and methodological standards. , 1985, The New England journal of medicine.

[20]  L. V. Hutton.  Using statistics to assess the performance of neural network classifers , 1992 .

[21]  Charles T. Butler,et al.  Neural networks, theory and practice , 1996 .

[22]  Jacek M. Zurada,et al.  Introduction to artificial neural systems , 1992 .

[23]  A. Detsky,et al.  Neural networks: what are they? , 1991, Annals of internal medicine.

[24]  James L. McClelland.  Explorations In Parallel Distributed Processing , 1988 .

[25]  Murray Smith,et al.  Neural Networks for Statistical Modeling , 1993 .

[26]  P. Wilding,et al.  The application of backpropagation neural networks to problems in pathology and laboratory medicine. , 1992, Archives of pathology & laboratory medicine.

[27]  Geoffrey E. Hinton.  Connectionist Learning Procedures , 1989, Artif. Intell..

[28]  R C Quijano,et al.  Neural net-bootstrap hybrid methods for prediction of complications in patients implanted with artificial heart valves. , 1994, The Journal of heart valve disease.

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

[30]  R. Simon,et al.  Flexible regression models with cubic splines. , 1989, Statistics in medicine.

[31]  R. Tibshirani,et al.  An introduction to the bootstrap , 1993 .

[32]  G S Doig,et al.  Modeling mortality in the intensive care unit: comparing the performance of a back-propagation, associative-learning neural network with multivariate logistic regression. , 1993, Proceedings. Symposium on Computer Applications in Medical Care.