Abstract:
In recent decades, and to this day computer technology has been used in applications and
various fields including the medical field, which prompted many researchers to employ this technique
in the design of decision support systems using many of the algorithms and methods for this purpose.
In this paper, k-means and back propagation are proposed to classify medical datasets and then
compare the performance of these methods, practical experiments show back propagation has best
results than k-means.

1 Introduction:
Data mining seeks a solution for real world health problems in the diagnosis
and handling diseases. Several data mining techniques were used by Researchers in
the medical field such as decision tree, k-means, fuzzy c-means, k-nn and neural
network(Das, 2009).
To decrease cost and human effects K.Rajalakshmi, Dr.S.S.Dhenakaran and
N.Roobini proposed a prediction system. This system includes analyze different
sickness prediction by using k-means clustering algorithm, for this purpose three
medical dataset were used (Heart Disease, Diabetics, Liver disease and
Cancer)(Rajalakshmi., 2015).
Nitu M. and Ashish B. implemented a classifier system containing k-means
and back propagation algorithms. They conducted a study by applying two methods
on the staffing data of an organization to analyze the performance of each method, at
end the result of study show that back propagation is better than k-means(Nitu, 2012).
Heart disease is the most common factor for death in India, in order to reduce
the risk of this disease, the trend has been to design decision support systems to help
doctors diagnose heart disease process with less features. From this standpoint, the
researchers Priti , M. A. jabbar and B.L Deekshatulua Chandra went to design the
system of diagnosis heart disease based on k-nn and genetic algorithm. They use
k-nn as a classifier method and genetic algorithm for reduce the features. The result of
system proves high accuracy(Priti, 2013).