Open AccessDiagnosing Diabetes with Extended Binary Cuckoo Search and k-Nearest Neighbor Classifier
Author(s) -
Shaymaa Abdul Hussein Shnain,
Zahraa Modher Nabat,
May A. Salih,
Baydaa Jaffer Al Khafaji
Publication year - 2020
Publication title -
xi'nan jiaotong daxue xuebao
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.308
H-Index - 21
ISSN - 0258-2724
DOI - 10.35741/issn.0258-2724.55.2.36
Subject(s) - classifier (uml) , cuckoo search , diabetes mellitus , artificial intelligence , computer science , k nearest neighbors algorithm , disease , binary classification , machine learning , pattern recognition (psychology) , data mining , medicine , support vector machine , pathology , particle swarm optimization , endocrinology
Diabetes is a common disease that develops at different ages. Sometimes the body of an individual who has diabetes either doesn’t need insulin or is resistant against insulin. Today, medical scientists and doctors face large amount of data. Since disease diagnosis is not simple work, in order to make a suitable decision, the doctor should investigate the patients’ tests’ results and the decisions that have been made for patients with the same status before. Using data-mining methods can help the early diagnosis of diabetes, which helps prevent this disease and a lot of its complications, such as cardiovascular disease, vision problems and nephrogenic disease. In this approach, k-nearest neighbor classification is used for classification and extended binary cuckoo search uses UC Irvine Machine Learning Repository UCI learning storage to select the diagnosis features of diabetes disease in a dataset of diabetes diseases. For classifying diabetes diseases, the research provides a system of diagnosis by utilizing the binary cuckoo search-optimized rough collections-based reduction of features and the classifier of k-nearest neighbor.