Design of a data analytics model for efficient prediction and monitoring of diabetes patients

Abstract

Diabetes is a condition of elevated blood glucose levels that newlineoccurs when the body can t produce or consume enough insulin. It is a newlineNon-Communicable Disease (NCD). However, according to the World Health newlineOrganization (WHO) 2016 report, the number of people living with diabetes newlineand its prevalence are increasing in all regions of the world as a result of newlinechanging eating habits and sedentary lifestyle. In near future, the number of newlineadults and even children suffering from diabetes is likely to increase newlinetremendously. The complications of diabetes can lead to heart attack, stroke, newlineblindness, kidney failure, and lower limb amputation. The task of this newlinediscipline is to analyse the diabetes data and develop strategies to reduce the newlinegrowing numbers in South East Asia and especially India.. newlineThe research work is focused on efficient prediction of diabetes newlineusing data analytics. The research work is based on disease pattern analysis, newlinerisk factor identification, glycemic load, visualization, and formulation newlinestrategies to prevent diabetes mellitus in India. The general framework of the newlineresearch work is provided in four divisions. The first is a data store that newlineprovides modalities to collect the data from various sources that would pave newlinethe research direction. The information was gathered primarily from the UCI newlinerepository, Twitter, and clinical trial data from hospitals. The pre-processing newlinesection extracts the data from the source, cleanses the data, and, if necessary, newlinetransforms them in order to impose computational models on them. For newlineexample, the data is label encoded to perform rank correlation. The data is newlineconverted to a Time Series for analysis and forecasting the time-based values newlineof critical variables. The computational models section renders the data over newlineseveral models for descriptive, predictive, and prescriptive analysis. Machine newlineLearning, Data Analytics, transfer learning and Deep Learning techniques are newlineused for the manifestation of data in time and space. The models are finally newlineevaluated with model-specific evaluation parameters newline