Copyright © 2018 Deepthi Gurram, M. R. Narasinga Rao. This is an open access article distributed under the Creative Commons Attribution
License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
International Journal of Engineering & Technology, 7 (1.1) (2018) 326-328
International Journal of Engineering & Technology
Website: www.sciencepubco.com/index.php/IJET
Research paper
A comparative study of support vector machine and logistic
regression for the diagnosis of thyroid dysfunction
Deepthi Gurram
1
*, M. R. Narasinga Rao
2
1
Research Scholar, Department of Computer Science, K L University, Vaddeswaram, Andhra Pradesh, India
2
Department of Computer Science & Engineering , K L University, Vaddeswaram, Andhra Pradesh, India
*Corresponding author E-mail: deepug2110@gmail.com
Abstract
Thyroid is one of the vital diseases that influence individuals of any age group now a day. Infections of the thyroid, incorporate condi-
tions related with extreme release of thyroid hormones (Hyper thyroidism) which is likewise called thyrotoxicosis and those related with
thyroid hormone insufficiency (Hypothyroidism). Expectation of these two sorts of thyroid disease is critical for thyroid analysis. In this
paper, support vector machines and logistic regression are proposed for predicting patients with thyrotoxicosis and without thyrotoxicosis.
The outcomes demonstrate that, logistic regression perform well over support vector machine with 98.92% exactness.
Keywords: Logistic Regression; Precision; Recall; Support Vector Machine; Thyrotoxicosis.
1. Introduction
These days, by the rapid development of innovation and infor-
mation in medical sciences, the software engineering experts are
fit for providing expert frameworks to determine various types of
diseases with high exactness. The therapeutic experts are made to
utilize these frameworks because of the existence of some prob-
lems at the time of prediction process [1]. Disease diagnosis oper-
ation utilizing expert frameworks are performed in light of a set of
disease manifestations. These frameworks depend on machine
learning procedures which help the doctor and a physician to limit
the expenses and time and can act as an expert advisor in making
successful conclusions.
In the human body, the thyroid organ is an essential organ. It pro-
duces thyroid hormones to keep up our body metabolism [2]. The
thyroid organ is situated in the front of the neck and underneath
the Adam's apple. The thyroid produces two noteworthy hormones
called T3 (triiodothyronine) and T4 (thyroxine). These T3 and T4
hormones make a trip in our blood to all parts of our body and
influence practically every cell in the body, and controls our
body's function. On the off chance that, the measure of thyroid
hormone diminishes in our blood and our body work gets back off,
this condition is called hypothyroidism [3]. The signs of hypothy-
roid are depression, exhaustion of body strength, tiredness, consti-
pation, excess weight, cramps, dry skin, sexual disorders and in-
fertility. On the off chance that, the increased measure of thyroid
hormones found in our blood, our body functions will increase.
This condition is called hyperthyroidism. The indications of hy-
perthyroid are anxiety, palpitation, exhausted body quality, trem-
ors, loss of weight, diarrhoea, menstrual disorder and exophthal-
mia. Specialists can fuse various factors, including clinical as-
sessment, blood tests, imaging tests, biopsies, and different tests to
analyze thyroid illness. A typical utilized strategy is a test, called
the thyroid- stimulating hormone (TSH) test, which can recognize
thyroid issue even before the onset of indications.
These days, CAD frameworks are getting increasingly prominent.
With the assistance of the CAD frameworks, the identified errors a
doctor can make, over the span of analysis can be kept away from,
and the medical information can be inspected in shorter time and
more definite as well [4]. Machine learning procedures are pro-
gressively acquainted with to build the CAD frameworks inferable
from its firm capacity of draw out complex relationships in the
biomedical information. As of late, different strategies have been
proposed to take care of this issue.
2. Related work
In 2014, Baydaa S. B. Alyas [5] planned an inventive framework
which would diagnose the thyroid patients with least execution
time and superior. This system should enable the healthcare ex-
perts to answer inquiries that describe the endocrine organ dys-
function and that allow them to take any clinical conclusions.
In 2013, Ahmad Taher Azar et al. [6] proposed a correlation
amongst hard and fuzzy clustering algorithms for thyroid maladies
informational index in order to locate the optimal number of clus-
ters. Distinctive scalar validity measures are utilized in looking at
the overall execution of the proposed clustering frameworks. To
locate the optimal number of clusters, elbow paradigm is enforced.
The clustering outcomes for all algorithms are then imagined by
the Sammon mapping strategy to locate a low-dimensional (ordi-
narily 2D or 3D) portrayal of a collection of points scattered in a
high dimensional pattern space.
In 2013, Ms. Wrushali Mendre Dr. Ranjana D. Raut [7] explored
the potentiality of neural network to segregate the two subtypes,
hypothyroid and negative form of thyroid issue based on the ar-
gument of laboratory medical information base. The best parame-
ters are distinguished for the neural networks like Multilayer Per-
ceptron (MLP), Radial Basis Function (RBF) and Principal Com-
ponent Analysis (PCA).
In 2012, Hui-Ling Chen, et al. [8] represented a three-phase expert
framework based on a hybrid support vector machines (SVM)