Volume 7, Issue 8, August – 2022 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Disease Prediction using Machine Learning

Anika Shreya Pawar
Dept. of CSE, BNMIT Bangalore, India

Abstract:- Machine learning (ML) is an artificial individual much better. Some of the domains of healthcare
intelligence (AI) technique that facilitates the where machine learning is being used are cancer detection,
improvement of predictability in software applications laser treatment, heart disease prediction, liver disease
without requiring explicit programming. Data from the prediction, kidney disease prediction and whether the
past is used to predict new outcomes using machine person is diabetic or not.
learning algorithms. During the course of this paper, we
used four machine learning algorithms: Logistic II. DATASET
Regression, Support Vector Machine, Decision Tree,
and Gradient Boosting. Our chosen algorithms were In this paper, five datasets are being used, each for one
applied to five datasets from the healthcare domain, in disease prediction. These datasets are available with no cost
which we were able to predict kidney disease, liver on the Kaggle website.
disease, breast cancer predictions, heart diseases, and A. Liver Disease Prediction Dataset
diabetes predictions. The liver dataset contains 11 attributes and 583 rows of
Keywords:- Machine Learning; kidney; heart; liver; data. This dataset contains column like the gender of the
breast- cancer; diabetes; svm; linear regression; decision patient, their age and other attributes like insulin level,
tree; gradient boosting. protein intake just to name a few. To predict whether the
patient has a liver disease or no there is a column attribute
I. INTRODUCTION with the title ‘Dataset’. This attribute is used in the
prediction process.
Machine learning can be defined as the use of
statistical models and probabilistic algorithms to answer B. Diabetes Prediction Dataset
questions so we can make informative decisions based on The diabetes dataset contains 9 attributes and 768 rows
our data. Machines make all these things possible by of data. This dataset contains column like pregnancies,
filtering useful pieces of information and piecing them glucose, blood pressure, skin thickness, insulin, BMI, age
together based on patterns to get accurate results. The and outcome. To predict whether the patient has diabetes or
different machine learning algorithms are supervised, no there is a column attribute with the title ‘outcome’.
unsupervised and reinforcement learning algorithms. In
supervised learning, we use known or labeled data for the C. Kidney Disease Prediction Dataset
training data. Since the data is known, the learning is, The kidney dataset contains 26 attributes and 400 rows
therefore, supervised, i.e., directed into successful of data. This dataset contains column like age, blood
execution. Two types of supervised learning algorithms are pressure, sugar levels, red blood cells, just to name a few.
regression and classification. Examples of this division of To predict whether the patient has a kidney disease or no
algorithm include decision tree, linear regression logistic there is a column attribute with the title ‘class’, which
regression, and support vector machine, just to name a few. defines the class of the disease.
The second type of machine learning algorithm is the D. Breast Cancer Prediction Dataset
unsupervised machine learning algorithm. In unsupervised The breast cancer dataset contains 33 attributes and 570
learning, the training data is unknown and unlabeled - rows of data. This dataset contains column that describe the
meaning that no one has looked at the data before. size, shape and location of the cancer. To predict whether
Clustering is a type of unsupervised learning algorithm. the patient is suffering from breast cancer or no there is a
Examples of this division of algorithm include singular column attribute with the title ‘diagnosis’, which describes
value decomposition, K-means clustering, Apriori, whether thecancer is benign or malignant.
Hierarchical clustering, Principal component analysis, just
to name a few. The third type of machine learning E. Heart Disease Prediction Dataset
algorithm is the reinforcement learning. The algorithm The heart dataset contains 14 attributes and 303 rows of
discovers data through a process of trial and error and then data. This dataset contains column like age, gender, chest
decides what action results in higher rewards. The agent, pain type, etc., just to name a few. To predict whether the
the environment, and the actions are the three major patient has a heart disease or no there is a column attribute
components that make up reinforcement learning. with the title ‘output’.
Examples of this division of algorithm include Q-learning,
R-learning and TD-learning.

Machine learning helps the healthcare specialists like

doctors and physicians to identify the healthcare needs
and solutions much faster and more accurately. Such early
and accurate predictions help in making the health of

IJISRT22AUG1127 www.ijisrt.com 1422

Volume 7, Issue 8, August – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
III. METHODOLOGY B. Support Vector Machine
Support Vector Machine (SVM) is one of the types of
We proposed a methodology that contains the supervised machine learning algorithm. It can be used for
following steps data collection, data pre-processing, both regression and classification problems, but it is
exploratory data analysis, feature selection, model selection primarily used for classification problems in machine
and prediction. First step is data collection. In this step, the learning. In SVM algorithms, the main goal is to create the
five different datasets are collected and then the data is best line or decision boundary that will segregate n-
loaded. The second step is data pre-processing step. In this dimensional space into classes so that it is easy to
step, the data that was collected is pre-processed like categorize a new data point. The best decision boundary is
checking for null values, checking for duplicate values and called a hyperplane. Some domains where SVM can be
removing them and dropping the unnecessary columns. applied are image classification, text categorization and
The third step is exploratory data analysis; basically it face detection.
refers to understanding the relationship between different
columns of the dataset. Now fourth step is feature selection, C. Decision Tree Classifier
it refers to the variable or attributes selection. In this step, Decision Tree is a supervised learning algorithm which
after selecting the most significant features we split the is preferably used for classification problems when
dataset into test and train data. The next step is model compared to regression problems. It has a tree like structure
selection. In this step, we select a model or a group of which follows the top-down approach. This decision tree
models that will predict the disease for us. The last step is consists of internals nodes which represent the features of
prediction. Prediction refers to predict the output after the the dataset, each leaf node represent a particular outcome
model has been trained on the trained data and apply to the and branches of the tree which represents the decision or
test data to observe the prediction. The figure (figure 1) rules. A decision tree simply asks a question, and based on
the answer (Yes/No), it further split the tree into sub trees.

D. Gradient Boosting Classifier

Gradient boosting classifier is a supervised learning
algorithm that is used to combine multiple weak models
which will help in strengthening the learner. One of the
main advantages of gradient boosting machines comes from
the fact that they can be used on more on multi-
classification problems than on binary classification
problems.

IV. EXPERIMENTAL RESULTS

We applied Logistic Regression, Support Vector

Machine, Decision Tree and Gradient Boosting Classifier
on all the five chosen datasets of the liver, kidney, heart,
breast-cancer and diabetes. The following results were
observed individually:

A. Experimental results for the Breast Cancer Prediction

below describes the architecture of the proposed system. Table 1: Results for Breast Cancer Prediction
Fig. 1: System Architecture As shown above, Gradient boosting classifier and
support vector machine showed the highest accuracy score.
A. Logistic Regression
Logistic regression is one of the most popular machine
learning algorithms. It is a type of supervised learning
algorithm. It is used for predicting dependent variables
using a set of independent variables. It predicts the output of
categorical dependent data. These categorical values could
be ‘Yes’ or ‘no’, or ‘true’ or ‘false’, or a probabilistic value
between 0 and 1. Logistic regression is a classification
based algorithm.

IJISRT22AUG1127 www.ijisrt.com 1423

Volume 7, Issue 8, August – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
As shown above, Decision Tree Classifier and
Gradient Boosting Classifier showed the highest accuracy.

Fig. 4: Correlation Matrix of Kidney Disease

D. Experimental results for the Diabetes Prediction

Fig. 2: Correlation Matrix of Breast Cancer

B. Experimental results for the Heart Disease Prediction

Table 4: Results for Diabetes prediction

As shown above, Gradient Boosting Classifier showed

thehighest accuracy.
Table 2: Results for heart disease prediction

As shown above, Logistic Regression algorithm

showed thehighest accuracy.

Figure 5: Correlation Matrix of Diabetes

E. Experimental results for the Liver Disease Prediction

Fig. 3: Correlation Matrix of Heart disease

C. Experimental results for the Kidney Disease Prediction

Table 5: Results for liver disease prediction

Table 3: Results for Kidney disease prediction

IJISRT22AUG1127 www.ijisrt.com 1424

Volume 7, Issue 8, August – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
REFERENCES
As shown above, Support Vector Machine and Gradient
Boosting Classifier showed the highest accuracy. [1.] DEEPTI, S. AND DILIP S. (2018). PREDICTION OF
DIABETES USING CLASSIFICATION
Fig. 6: Correlation Matrix of Liver disease ALGORITHMS. INTERNATIONAL CONFERENCE
ON COMPUTATIONAL INTELLIGENCE AND
F. Comparing the results Disease Prediction DATA SCIENCE.
[2.] FANIQUL ISLAM M.M ET AL. (2020)
'LIKELIHOOD PREDICTION OF DIABETES AT
EARLY STAGE USING DATA MINING
TECHNIQUES.' COMPUTER VISION AND
MACHINE INTELLIGENCE IN MEDICAL IMAGE
ANALYSIS. SPRINGER, SINGAPORE, 113-125.
[3.] HASSAN A.S ET AL. (2020). DIABETES
MELLITUS PREDICTION USING
CLASSIFICATION TECHNIQUES.
INTERNATIONAL JOURNAL OF INNOVATIVE
TECHNOLOGY AND EXPLORING ENGINEERING
(IJITEE) ISSN: 2278-3075, VOLUME-9 ISSUE-5.
[4.] MAHBOOB T. A ET AL. (2018). A MODEL FOR
EARLY PREDICTION OF DIABETES.
INFORMATICS IN MEDICINE UNLOCKED.
[5.] “PERFORMANCE EVALUATION OF MACHINE
LEARNING METHODS FOR BREAST CANCER
Table 5: Comparitive results for disease prediction PREDICTION”, BY YIXUAN LI, ZIXUAN CHEN
OCTOBER 18, 2018.
V. CONCLUSION [6.] “BREAST CANCER PREDICTION AND
DETECTION USING DATA MINING
If the healthcare related diseases are diagnosed at their CLASSIFICATION ALGORITHMS: A
COMPARATIVE STUDY” BY MUMINE KAYA
KELES, FEB 2019.
[7.] “BREAST CANCER PREDICTION USING DATA
MINING METHOD ” BY HAIFENG WANG AND
SANG WON YOON, DEPARTMENT OF SYSTEMS
SCIENCE AND INDUSTRIAL ENGINEERING
STATE UNIVERSITY OF NEW YORK AT
BINGHAMTON BINGHAMTON, MAY 2015.
early stages, this will help thousands of people to save [8.] “MACHINE LEARNING WITH APPLICATIONS IN
themselves and take the right medication at an early stage to BREAST CANCER DIAGNOSIS AND
stop the disease from becoming worst. We may be capable PROGNOSIS” BY WENBIN YUE, ZIDONG WANG,
of classifying and are expecting the most disease to be 9 MAY 2018.
classified as either a one at an early stage or the ones at an [9.] M. CHEN, Y. HAO, K. HWANG, L. WANG, AND L.
final stages by making use of the machine learning WANG, “DISEASE PREDICTION USING
algorithms. Machine Learning algorithms may be used for MACHINE LEARNING OVER BIG DATA.” IEEE
scientific orientated studies, it speed up’s the system, ACCESS ( VOLUME: 5), PP. 8869 – 8879.
reduces the human mistakes & decreases the guide mistakes. [10.]Z. CHEN, Z. ZHANG, R. ZHU, Y. XIANG AND P.
B. HARRINGTON, "DIAGNOSIS OF PATIENTS
VI. FUTURE ENHANCEMENT WITH CHRONIC KIDNEY DISEASE BY
USING TWO FUZZY CLASSIFIERS",
Many more Machine Learning algorithms can be
CHEMOMETRICS INTELL. LAB. SYST., VOL. 153,
used to predict a much larger number of diseases. A variety
PP. 140-145, APR. 2016.
of algorithms can be combined to get better predictions.

Deep Learning techniques can be made use of in the

healthcare sector to help in predicting and classifying
diseases.

IJISRT22AUG1127 www.ijisrt.com 1425