International Journal of Scientific Research in Computer Science, Engineering and Information Technology
© 2019 IJSRCSEIT | Volume 5 | Issue 2 | ISSN : 2456-3307
DOI : https://doi.org/10.32628/CSEIT1952189
Disease Prediction Application based on Symptoms
Ramandeep Singh Sethi1, Aniket Thumar1, Vaibhav Jain1, Sachin Chavan2
Student, Computer Department, NMIMS Shirpur, District: Dhule, Maharashtra, India
1
2Professor, Computer Department, NMIMS Shirpur, District: Dhule, Maharashtra, India
ABSTRACT
The usage of mobile phones in today’s world is more than ever. Mobile phones are everywhere and the mobile
technology is growing at an exponential rate. The capabilities of a mobile phone have made it provide us services
that make human life better. One such service that mobile phones can offer us is digital healthcare. Also, it is
recognized that mobile phone applications that provide healthcare solutions are trending. Such applications
provide a convenient and portable healthcare solutions to all the individuals. Such applications provide a rich
experience to a user and in this way, the users will come to know more about their health and body. Digital
healthcare mobile applications are capable of diagnosing a disease that a patient is suffering from using his/her
symptoms. This information can be used further by a medical practitioner for later on consultation.
Keywords : Data Mining, Naïve Bayes Classifier
I. INTRODUCTION II. LITEARTURE WORK
There are several disease predication systems available Darcy A. Davis Used ICD9-CM to predict future
in market which are available to medical practitioners disease risks. They used clustering to predict the
only. Such systems have decent accuracy but an end disease based on similar patient’s medical history [1].
user cannot use it on his/her mobile device. Other T.F. Michael Raj and S. Prasanna proposed the model
web-based systems are available to end users but these that trains the machine and it proves the probabilistic
systems are not portable (Cannot use them as a mobile models are stable and reliable to identify the disease
application) and they do not provide a personalized [2]. K.Rajalakshmi, Dr.S.S.Dhenakaran, N.Roobini
treatment and remedies. To increase the prediction pre-processed data collected from different sources
accuracy, the application must also take care of the was given as input to the different clustering methods.
patient’s BMI (Body Mass Index) into account while When K-Means algorithm was applied to preprocessed
prediction. There is a lot of medical data available data, it showed low accuracy [3]. But when it was used
which can be used to predict the disease of a patient along with different classifiers, it showed decent
based on the symptoms that he/she enters in the accuracy [4].Shalet K.S, V.Sabarinathan,
application. Several other parameters like age, gender, V.Sugumaran, V.J. Sarath Kumar proposed a model
weight and height play a crucial role in disease where REPTree was used for the process of feature
prediction. The mobile application should also selection. This selection helped us get structural
recommend the near-by hospitals where the information which can be analyzed easily. Feature
diagnosed disease can be predicted. classification is done using SVM (support vector
machine) [5]. Some authors have conducted the
CSEIT1952189 | Received : 10 March 2019 | Accepted : 24 March 2019 | March-April -2019 [ 5 (2) : 641-646 ]
641
Ramandeep Singh Sethi et al Int J Sci Res CSE & IT. March-April-2019 ; 5(2) : 641-646
research for the best medical diagnosis mining • P(c | x) is the probability of occurence (c,
algorithm. These authors compared Naïve Bayes with target) given (x, attributes) has occured.
five other classifiers i.e. Logistic Regression, K-Star, • P(c) is the probability of occurrence of class c.
Decision Tree, Neural Network and a simple rule- • P(x | c) is the probability for x given that c has
based algorithm [6]. For evaluating the performance of already occured.
all algorithms, 15 real-world medical problems from • P(x) is the probability of event x.
the UCI machine learning repository were selected. In
the experiment, it was found that Naïve Bayes defeats The patient is allowed to create an account and he/she
the other algorithms in 8 out of 15 data sets [7]. So, it can login using his/her credentials. The login page is
was concluded that the predictive accuracy results in designed as shown below:
Naïve Bayes is better than other techniques [8].
Figure 3.1 Login Screen
Figure 2.1. Flowchart of proposed system
To signup, the patient has to choose a username,
III. IMPLEMENTATION password and enter his age, weight, gender and height.
Based on our study, Naïve Bayes algorithm has proven The sign-up page is shown as below:
to be the best for disease prediction systems.
Bayes theorem provides a way of calculating posterior
probability P(c|x) from P(c), P(x) and P(x|c):
𝑃(𝑥|𝑐)𝑃(𝑐)
𝑃(𝑐|𝑥) =
𝑃(𝑥)
• P(c | X) =P(x1|c) * ……* P(xn | c)*P(c)
Volume 5, Issue 2, March-April -2019 | http://ijsrcseit.com
642
Ramandeep Singh Sethi et al Int J Sci Res CSE & IT. March-April-2019 ; 5(2) : 641-646
Figure 3.2 Signup Screen
Figure 3.4 Home screen
A general health background of patient is also
necessary for predicting the right disease. So, during
the signup process, the patient has to answer some
questions related to his health background:
Figure 3.5 Profile screen
Figure 3.3 Health Background Screen
Volume 5, Issue 2, March-April -2019 | http://ijsrcseit.com
643
Ramandeep Singh Sethi et al Int J Sci Res CSE & IT. March-April-2019 ; 5(2) : 641-646
Figure 3.6 Questionnaire start screen Figure 3.8 Result screen
Figure 3.9 Report screen
Figure 3.7 Questionnaire screen
Volume 5, Issue 2, March-April -2019 | http://ijsrcseit.com
644
Ramandeep Singh Sethi et al Int J Sci Res CSE & IT. March-April-2019 ; 5(2) : 641-646
Figure 4.1 Result analysis
V. CONCLUSION
Here we have implemented an improved approach for
generalized disease prediction system based on
symptoms by using Naïve Bayes algorithm. Our
system takes one-time input such as Age, Height,
Figure 3.10 Nearby hospitals screen
Weight, Gender, etc. from the user and this data is
used further in disease prediction. Our system asks
IV. RESULT ANALYSIS
user some simple & relevant questions and based on
the answer given by the user, it generates a report
The app was tested with 200 patients. Patients had to
which contains list of diseases with their probability
start symptom assessment, record the results of app
and list of nearby hospitals. The report contains the
and visit a medical practitioner to know the actual
causes & remedies of the disease with maximum
disease. These details were submitted by them in a
probability. User can also update his personal
survey (Google Form). The accuracy of present
information (Age, Weight, Height, and Password). It
healthcare system (ADA) is 34% when top 5 diseases
is our opinion that research in healthcare is an exciting
predicted are taken into consideration. Our system is
area for many years to come and will keep many
accurate 42% of the times that is, the correct disease
scientists and engineers busy.
(top 3 predictions) was predicted in 84 patients out of
200.
VI. FUTURE WORK
In the future, our system can be extended by
deploying an additional feature of early prediction,
where the system will notify the user about a possible
disease before the user even starts a questionnaire.
Also, our mobile application can be integrated with
healthcare bands and rings available in the market for
better predictions. This will also provide a great
healthcare experience to the user.
Volume 5, Issue 2, March-April -2019 | http://ijsrcseit.com
645
Ramandeep Singh Sethi et al Int J Sci Res CSE & IT. March-April-2019 ; 5(2) : 641-646
REFERENCES
[1]. Davis, D., V. Chawla, N., Blumm, N., Christakis,
N., & Barbasi, A. L. (2008) “Predicting Individual
Disease Risk Based on Medical History”
[2]. Adam, S., & Parveen, A. (2012) “Prediction
System for Heart Disease Using Naive Bayes”
[3]. K.M. Al-Aidaroos, A. B. (n.d.). 2012. “Medical
Data Classification with Naive Bayes Approach”
[4]. Purvashi Mahajan, Abhishek Sharma “Role of K-
Means Algorithm in Disease Prediction”
[5]. K.Rajalakshmi, Dr.S.S.Dhenakaran and
N.Roobini “Comparative Analysis of K-Means
Algorithm in Disease Prediction”
[6]. Shalet K.S, V.Sabarinathan, V.Sugumaran, V.J.
Sarath Kumar “Diagnosis of Heart Disease Using
Decision Tree and SVM Classifier”
[7]. T.F. Michael Raj and S. Prasanna
“Implementation of ML Using Naīve Bayes
Algorithm for Identifying Disease-Treatment
Relation in Bio-Science Text”
[8]. Kazmierska J, Malicki J. “Application of the
Naïve Bayesian Classifier to optimize treatment
decisions”
[9]. Bairstow PJ, Mendelson R, Nguyen L. “Reducing
inappropriate diagnostic practice through
education and decision support”
Cite this article as :
Ramandeep Singh Sethi, Aniket Thumar, Vaibhav Jain,
Sachin Chavan, "Disease Prediction Application based
on Symptoms", International Journal of Scientific
Research in Computer Science, Engineering and
Information Technology (IJSRCSEIT), ISSN : 2456-
3307, Volume 5 Issue 2, pp. 641-646, March-April
2019. Available at doi :
https://doi.org/10.32628/CSEIT1952189
Journal URL : http://ijsrcseit.com/CSEIT1952189
Volume 5, Issue 2, March-April -2019 | http://ijsrcseit.com
646
