1
Liver Disease Prediction Using Machine Learning
Abstract- Liver disease had became one of the most
prominent disease in our country .It is the reason for about
2.4% of death per year in India. It has became a challenge to
predict the disease of liver in early stage if not diagnosed
early stages it become very hard to cure later on. Machine
Learning has helped us a lot in the field of medical.In this
paper,it is estimated that which attributes are important and
which are not.The classification techinques are performed
in the training dataset.The main aim of the paper is to apply
various machine learning algorithms like random
forest,SVM(Support Vector Machine),Logistic Regression
on the datasets and thus identify whether the patient has liver
disease or not.
Index Terms-Liver disease,classification,dataset,random
forest,SVM,logistic regression.
I. INTRODUCTION
Internet is full of data and people give their opinion about
everything and anything.It is extremely difficult to make a
correct choice when their is a large amount of data and no
good method to make a decision.We need a method to
extract the sentiment out of a data and use it as a make a
sensible choice , to solve such problem we use sentiment
analysis.SENTIMENTAL ANALYSIS is a kind of text
classification based on Sentimental Orientation (SO) of
opinion they contain. Sentiment analysis of product reviews
has recently become very popular in text mining and
computational linguistics research.
DISEASE OF LIVER
The main reasons of the liver disease are excessive
consumption of alcohol,fat accumulation in the liver,etc.
The common liver disorders are:
● Alcoholic hepatitis – This is swelling of the liver.
It might be mild and last for years but with no
certain symptoms.In mild cases,it might be
controlled by reducing the consumption of alcohol.
If not controlled,it might become fatal in that stage.
● Fatty liver disease-In this disease,fat is accumulated
in the liver cells.No symptoms are shown in this
disease.Most probably,it goes away when a person
stop drinking alcohol.
● Cirrhosis-It generally occurs when healthy liver
tissue is replaced by hard scar tissue.It is the most
serious form of liver disease.It stop liver from from
functioning properly and person may suffer from
liver cancer.It cannot be reversed but can be stopped
if the consumption of alcohol is stopped.
II. LITERATURE SURVEY
This part consist of papers that are surveyed:
P.Rajeswari,G.Sophia Reena et al.,[2010]introduced the
classification based on liver diagnosis. The dataset for
training is created by collecting data from UCI repository
which consists of 345 instances and 7 distinct attributes.
In this paper,the result is obtained by applying naïve baised
algorithm,K-star algorithm and FT tree algorithm.Out of
these three algorithms,time taken by FT tree algorithm is fast
with accuracy of 97.10%.Based on the result,FT tree
algorithm is the best algorithm among all three
algorithms[2].
Sa’diyah Noor Novita Alfisahrin, Teddy Mantoro et al.,
[2013] have introduced to predict or identify if the patients
is suffering from liver disease pr not on the basis of 10
attributes.The algorithms used in it were Naïve
Baiyes,Decision tree and NB tree algorithm.On seeing the
result,it is seen that NB Tree algorithm is the best algorithm
with the highest accuracy,however NB algorithm is the first
to give the results.For further study,the improvement in
accuracy of NB tree algorithm,will be the target by finding
the most suitable factors in predicting whether patient has
liver disease or not[3].
S.Dhamodharan[2014] introduced that there are many liver
disease which require medical attention.With the help of
different symptoms,3 major disease of liver are predicted
that are hepatitis,cirrhosis and liver cancer.The main aim of
the paper is to predict the type from classes like
cirrhosis,liver cancer,hepatitis or no disease.Algorithms
used in this were Naïve Bayes and FT tree which on
comparing the accuracy,it is seen that Naïve Bayes
algorithm is better than other algorithms[4].
A.S.AnneshKumar ,Dr.C.JothiVenkateswaran et al.
,[2015]has introduced which describes the categorization
of liver disease by use of fuzzy K-means classification and
feature selection. Various liver diseases have similar
attribute values and thus it require more hard work to
classify liver disease type correctly. So Fuzzy based
classification best performance in these classes and gave

above 94 percentage accuracy for every type of liver
disorder [6].
P.Thangarajul, R.Mehala et al., [2015] has introduced to
analyze liver diseases patients datawith the use of particle
swarm optimization algorithm (PSO) with K-Star
classification in two ways for classifying the presence of
disease or not.This algorithm increase the performance of
accuracy when compared to present classification
algorithms. PSO-Kstar algorithm is best algorithm for the
liver disorders classification as it improved the
performance in prediction accuracy.The best data mining
algorithm with respect to understandability,
transformability and accuracy is PSO-KStar algorithm with
100% accuracy[7].
Onwodi Gregory [2015] introduced two dataset of liver
patient which were to build classification model for
predicting liver disorder . Eleven datamining classification
algorithm were used in dataset and then performance of all
those algoritim are compared among themselves with
repect to accuracy , recall and precision. Based on those
result the accuracy of FT tree algorithum was the best
which 78% accuracy , 86.4% sensitivity ,77.5% of
precision and 38.2% of specificity result respectively [8].
III. IMPLEMENTATION
i)DATASET
The dataset of Indian Liver Patients comprises of 11
attributes and 583 liver patients.The patients in dataset were
given by either 1 or 2 for liver patient or not.
The dataset was records of the patients of liver from the
hospital of Andhara Pradesh.
The dataset comprises of the following attributes:
⮚ Age
⮚ Sex
⮚ Total_Bilirubin
⮚ Direct_Bilirubin
⮚ Alkaline Phosphatase
⮚ Alamine Phosphatase
⮚ Total_Protein
⮚ Albumin
⮚ Albumin and Globulin_Ratio
⮚ Result
ii)Data-preprocessing
Data-preprocessing means transforming the data before
using it .It is done to convert the raw data into clean data.
2
Due to data-preprocessing,it becomes easy to process the
data and thus gives the better result.The example of the data-
preprocessing is to fill the null values in the dataset.
The three techniques used in data-preprocessing are:
● Rescale data
● Binarize the data
● Standardize the data
iii)Machine learning Algorithms used:
A. SVM(Support Vector Machine)
This algorithm aims to focus on to find the hyperplane in N-
dimension plane which classifies the data-points distinctly.
Hyperplanes are the boundries which classifies the
datapoints distinctly.Various hyperplanes are created and
hyperplane with the largest margin is choosed in this
algorithms to classify.The more is the margin,the less is the
error.
B. Random Forest algorithm
It is the flexible,easy to use algorithm in the field of machine
learning.It is a supervised algorithm and can be used for both
classification and regression.Random forest reduce the
variance that create the disturbances in the results.We can
find the output of the individual tree through majority voting
and thus smoothing out the variance to increase the accuracy
of the results.
C. Linear Regression
This is a type of supervised algorithms.This algorithm
models a target precision on independent variable.It is called
so because it finds the linear relationship between X and Y
where X is input and Y is output.The best fit line for our
model is regression line.
D. Logistic Regrssion
It is another technique in the field of machine learning from
field of statistic.It is used in binary classification method.To
squash the value between 0 and 1,it makes use of logistic
function or sigmoid function which is a S shaped graph.
E. Naïve Bayes algorithm
Although this algorithm is simple but is very powerful
algorithm for predictive model in classification.This
classifier is from the family of probabilistic classifier which
makes use of the Bayes theorem with strong independence
assumption between features.It considers each feature to
 3

contribute independently to the probability regardless of any

possible correlation between the attributes. Name of the classification algorithm Accuracy
1. SVM 70.85%
2. Logistic Regression 71.42%
IV.RESULT AND ANALYSIS 3. Linear Regression 65.71%
4. Random forest 65.71%
In this section,different algorithms are compared to each 5. Naïve Bayes 58.85%
others on the basis of following factors: Result of algorithms

i)Accuracy
V.CONCLUSION
It is the ratio of the correct predicted to the total number of
predictions.
In this paper,various algorithms are applied on the
Formula of accuracy is:-
dataset.Various algorithm such as Random
forest,SVM,logistic regression,etc are applied to the
dataset.These algorithms gave different accuracy for the
dataset.
On analyzing the resul,it is seen that logistic regression
regression turns out to be the best algorithm with the
ii)Recall or sensitivity accuracy of 71.42%.

It is the ratio of the true positive to the sum of true

positive and false negative.
It can be calculated as:
Sensitivity = (True Positive)/(True Positive + False
Negative)

iii)Precision

It is the ratio between true positives and all positive

results.
It is given by:

Each algorithms give its own different result.

The result of each algorithm is shown in table below:
4
5