Diabetes Diagnosis with Maximum Covariance Weighted Resilience Back Propagation Procedure

This study presents Diabetes Diagnosis with Maximum Covariance Weighted Resilience Back Propagation Procedure. The Maximum covariance method is divided into three phases. A large number of candidate’s hidden units is considered by initializing their various weights with random values. Then the desired number of hidden units is selected amongst the candidates by using the maximum covariance. The weights feeding the output units are calculated with linear regression method. After the maximum covariance initialization, the network is trained with the resilient back propagation which is an adaptive training algorithm. The activation function in the hidden units is hyperbolic tangent function. Ten baseline variables includes, age, sex, body mass index, average blood pressure and six blood serum measurements, were obtained for each of n = 442 diabetes patients, as well as the response of interest, a quantitative measure of disease progression one year after baseline was used. The learning machine was trained, validated and tested. The result shows the algorithm is efficient in the diagnosis of who is a diabetic patient.