Title: Web-based multi-featured input diabetes mellitus diagnosis system using combined ANN-ARM techniques
Authors: K. Sridar; D. Shanthi
Addresses: Department of CSE, Veerammal Engineering College, K.Singarakottai, Tamil Nadu, India ' Department of CSE, PSNA College of Engg & Tech., Dindigul, Tamil Nadu, India
Abstract: Diabetes mellitus is a chronic disease that makes millions of lives miserable. The International Diabetes Federation survey states that 246 million people suffered from type I and type II diabetes. We have proposed a hybrid analysis, which takes real-time glucometer readings, clinical data as numeral or text inputs and retinal image (visual) as another input. The extracted features are trained, dataset formation and association analysis are done by Back Propagation algorithm (ANN) and Apriori algorithm (ARM). The same are analysed through hierarchical extreme learning machines (H-ELM) Neural Network (ANN) and FP-Growth (ARM) algorithms together. The proposed diagnosis system is implemented as a web-based application. The output of the system analyses the given inputs and shows the severity of the diabetes mellitus as type I, type II and type III. The patients themselves can identify their risk status. Thus the effective diagnosis of diabetes with better accuracy is achieved.
Keywords: diabetes mellitus; plasma glucose test; hierarchical extreme learning machines; FP-Growth; back propagation; apriori; retinal blindness; disease diagnosis.
International Journal of Biomedical Engineering and Technology, 2017 Vol.25 No.2/3/4, pp.300 - 309
Received: 28 Jan 2017
Accepted: 17 May 2017
Published online: 23 Oct 2017 *