Data Mining Driven Models for Diagnosis of Diabetes Mellitus: A Survey

B Z Yahaya,F S Ishaq,L J Muhammad,Y Atomsa

doi:10.17485/ijst/2018/v11i42/132665

Abstract

Objective: In this study, a systematic effort was employed to identify and review data mining concept, tasks and model evaluation techniques, Knowledge Discovery and Data mining process Model (KDDM) model process and research articles published with reputable journal publishers that employed data mining techniques for diagnosis of Diabetes Mellitus. Method/Analysis: The findings from this work have been drawn from the published articles reviewed and the frequency analysis was used for the analysis of the reviewed works. Finding: The result of the study showed that, classification data mining task has been the most successfully and most frequently used data mining tasks for diagnosis of DM and the mostly commonly used classification data mining algorithms are Support Vector Machine and decision tree algorithms. Novelty/Improvement: In the study Support Vector Machine was realized to be most efficient data mining algorithm for diagnosis of Diabetes Mellitus using either clinical or biological and clinical dataset of Diabetes Mellitus. Despite its popularity, SVM algorithm should be further improved in the future work so as to further improve its efficiency. Keywords: Algorithm, Data Mining, Diabetes; Diagnosis, Knowledge, Pattern

Highlights

Data mining is a process of uncovering hidden patterns or useful knowledge from large dataset or database
Support Vector Machine algorithm was used to identified the effectiveness of different types of treatment of diabetic patients for different age groups using dataset of non-communicable disease risk factors in Saudi Arabia which obtained from World Health Organization (WHO) in[6]
The model was proposed to improve the accuracy of the prediction of diabetes mellitus type II and result of the study showed that, the model attained a 3.04% higher accuracy of prediction compared to other researches in the work

Summary

Introduction

Data mining is a process of uncovering hidden patterns or useful knowledge from large dataset or database. As a result of its explosive rapid growth, data mining have turn out to be an increasingly vital research area[3]. It is becoming widely accepted in medical field, because of its efficient analytical techniques and algorithms, as it uncovers useful and valuable knowledge in medical datasets or databases. To open a window of comparatively better resources data mining techniques are employed to enhance the sensitivity and/or specificity of disease detection and diagnosis. This substantially reduces the accompanied cost as unwanted and expensive medical tests are bypassed[5]

Methods

Results

Conclusion