Developing A Diabetes Detection System

1585 WordsMar 22, 20167 Pages
Developing A Diabetes Detection System by Using Big Data Technology 4 Problem Solution In this system, the performance of CBR Algorithm will be boosted based on MapReduce approach and to detect diabetes of a particular patient with improved CBR algorithm by using Apache Hadoop framework. Fig. 1. Framework of Case Based Reasoning Algorithm The biggest challenge of a CBR is finding an accurate indexing function. In diabetes dataset, there are five variables that influence the glycosylated homeglobin level in blood. These are the stabilized glucose level, height, weight, waist, and hip sizes. It is medically known that whenever the level of GH in a person’s blood is over 7, the person is considered to have diabetes. These variables have various influences on GH thus they will have different weights in diabetes detection function. Moreover, obesity is a known factor for type 2 diabetes (T2D). Obesity of a person can be determined using body mass index (BMI). BMI is strongly and independently associated with the risk of being diagnosed with T2D. So, the best indexing formula is In diabetes detection system, the most challenging part about MapReduce is to adapt already existing functions to do the relevant system’s own tasks. There consists of two main blocks or classes which are map and reduce. In the map function, the data from Hadoop’s Distributed File System (HDFS) are divided using mapping tools into several parts. The next step is to analyze that data and
Open Document