Comparison of ANN models to predict LDL level in Diabetes Mellitus type 2

Abstract

Diabetes Mellitus type 2 (T2DM) is the disorder of hormone insulin. Abnormal insulin action increases the free fatty acid pool and favours the formation of low-density lipoprotein cholesterol (LDL-C). Increase in the level of LDL-C causes deposition of plaques in the vascular wall. This is responsible for vascular events like coronary vascular disease, peripheral vascular disease and neuropathies in T2DM patients. So, we need to monitor the levels of lipid profile of patients who are suffering from T2DM for long duration, to prevent them from peripheral as well as coronary vascular diseases. But a major part of the population of our country is below the poverty line and it is difficult for them to afford for the cost required for all these biochemical tests. In this paper, we presented two artificial neural network (ANN) based computational models: (1) back propagation-multilayer perceptron (BP-MLP) and (2) probabilistic neural network (PNN) for comparative study of prediction of LDL-C level. The predicted results were validated with clinical experimental results.