Title: A framework for predicting malaria using naïve Bayes classifier

Authors: Aminu Aliyu; Rajesh Prasad; Mathias Fonkam

Addresses: School of IT and Computing, American University of Nigeria, Yola, Adamawa State, Nigeria ' School of IT and Computing, American University of Nigeria, Yola, Adamawa State, Nigeria ' School of IT and Computing, American University of Nigeria, Yola, Adamawa State, Nigeria

Abstract: Malaria, a life-threatening parasite contained in the spittle of mosquitoes, and is transmitted via a bite. This study designs a framework for predicting malaria using a probabilistic classifier: naïve Bayes. The study classifies the incoming patient into two phases. In the first phase, it first classifies patients as either having malaria or not, then in the second phase it proceeds to further classify the level of severity. The framework has been tested on a sample dataset of 700 records obtained from a hospital located in Yola, Adamawa State of Nigeria. The result proved that the model can classify any given patients successfully, having provided the required input symptoms at both classification phases. The accuracy of the model was checked using confusion matrix and ROC.

Keywords: malaria; data mining; classification; naïve Bayes; prediction; performance measure.

DOI: 10.1504/IJTMCP.2018.093623

International Journal of Telemedicine and Clinical Practices, 2018 Vol.3 No.1, pp.78 - 93

Available online: 18 Jul 2018 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article