Volume no :16, Issue no: 1, September (2016)

ARTIFICIAL NEURAL NETWORKS AND LOGISTIC REGRESSION WITH AN ORDERED CATEGORICAL RESPONSE: STUDY CHRONIC KIDNEY DISEASE

Author's: Mona N. Abdel Bary, Dina H. Abdel Hady and Zenab M. EL Gamal
Pages: [67] - [91]
Received Date: August 19, 2016
Submitted by:
DOI: http://dx.doi.org/10.18642/jsata_7100121706

Abstract

The health research has become increasingly reliant on statistical modelling techniques to assess the effect of new health programs, the impact of risk factors on disease, the effects of health behaviours, and a host of other health concerns. Clinical researchers conduct studies about diagnostic tests mainly for the purpose of either estimating the diagnostic accuracy of a test according to different patient or environmental characteristics or comparing diagnostic accuracy of different tests according to different patient or environmental characteristics. Studies are required to develop robust statistical methods to analyze data from diagnostic studies and assess the properties of available statistical methods. Multiple analytical statistical methods are available to analyze ordinal data. These methods can be a model based approach, such as models for cumulative response probabilities or a non-model based approach, such as a nonparametric method based on ranking. A common model based method used to analyze ordinal data is an ordinal logistic regression. In addition to statistical models, several machine learning algorithms are also available to analyze ordinal data, such as an artificial neural network model, a decision tree model, and a support vector machine model. The current study compares the performance of ordinal logistic regression model with artificial neural networks models for prediction of chronic kidney disease. The results of the current study show outstanding performance of artificial neural networks models for the prediction of the level of chronic kidney disease. In addition, the study illustrates the most affected variables are gender, surgical operations, the blood pressure and potassium ratio. There are many previous studies focus on demonstrated the ability of ANN models in applications binary classifications but through the current study we use neural networks models in multiple classifications. The results of the current study show the following: successful of ANN models in the process of separating and classifying accuracy rate of almost , and the ordinal regression model succeeded in identifying risk factors for renal failure moral influence on the regression model.

Keywords

ordinal logistics regression model, artificial neural network, medical diagnosis, chronic kidney disease.