Comparison of artificial neural network and binary logistic regression for determination of impaired glucose tolerance/diabetes

A. Kazemnejad, Z. Batvandi, J. Faradmal
2010 Eastern Mediterranean Health Journal  
Discrimination between dietary specializations of bats has been largely analyzed using multivariate techniques such as discriminant and principal component analysis. In this study, models based on an artificial neural network (Multi-layer feed forward neural network) and Binary Logistic Regression (BLR) were compared in their ability to differentiate between insectivorous and frugivorous bats using habitat and morphometric measurements on captured bats. Although both models had similar
more » ... c performance based on the area under the ROC (99% vrs 99.09%), sensitivity (97.6% vrs 96.8%) and specificity (95.3% vrs 93.8%) values, the logistic model was superior to the neural network model. We therefore recommend that if prediction is the sole objective, then ANNs provide acceptable results whiles BLR could be used to identify factor effects on classification. Further studies on these models may consider incorporating other dietary habits as well as factor effects (predictors) which could improve the accuracy of predictions.
doi:10.26719/2010.16.6.615 fatcat:3ieqq5njmfgsjgqc2zspemfp5u