2022 International Conference on Machine Learning, Big Data, Cloud and Parallel Computing (COM-IT-CON), 26-27 May 2022

Special Disease Prediction System Using Machine

2022 International Conference on Machine Learning, Big Data, Cloud and Parallel Computing (COM-IT-CON) | 978-1-6654-9602-5/22/$31.00 ©2022 IEEE | DOI: 10.1109/COM-IT-CON54601.2022.9850843

Learning
Rohit Kumar, Prince Thakur, SPS Chauhan
School of Computing Science and Engineering, Galgotias University
Greater Noida, UP, India-201310
rohitraks583@gmail.com
princekumarth@gmail.com
sps.chauhan@galgotiasuniversity.edu.in

Abstract— The wide variety of computer-based technologies a robust tool for diagnosing diseases effectively.
within the healthcare industry has led to the gathering of
electronic data. Due to the massive number of information, consistent with our study general mathematical
medical professionals are faced with the challenge of accurately techniques, work experience and understanding of
diagnosing signs and figuring out diseases at an early stage. In medical professionals led to unpopular selections
medicine, misdiagnosis could be a major factor leading because
of poor treatment and diagnosing the disease when it’s serious. and errors when detecting disease-related risks.
However, supervised machine learning techniques have With the rapid proliferation of electronic health data,
demonstrated the potential to surpass conventional diagnostic medical doctors are facing the challenge of
procedures and assist medical professionals in diagnosing high-
risk diseases. Mostly people feel lazy to visit a hospital, and diagnosing disease accurately in advance. For this
concern a doctor for a minor complication. However, this small reason, advanced calculation methods such as
problem can pose significant medical risk. Since, online medical supervised ML algorithms are introduced to detect
advice is readily available. The system evaluates the symptoms
that person give as an input and gives the disease as an output. logical patterns and hidden information from data,
Naive Bayes Classifier is used in the system. Our system focuses which can be used to make critical decisions. As a
on accuracy, the more numbers of a symptoms furnished by the result, the burden on medical staff will be decreased,
person as a input the disorder prediction as a output will be
better. Work can enhance the health care industry to zenith and and the death rate of patients will be reduced and
give cure to world. survival rate will be enhanced.
.
Keywords— Supervised Machine Learning, Health care, Disease
Prediction, Naïve Bayes classifier II. PROPOSED SOLUTION

I. INTRODUCTION
As we all know Machine learning algorithms use
a range of developmental, mathematical, and
probable development techniques to learn from data Fig.1. Block Diagram
generated from past events, and use them in higher
cognitive process. The enhancement of AI has Classical process could be a procedure within which a
enabled computer systems to acknowledge, think patient should visit a doctor, receive various treatments and
and use intelligently as human beings. sML are at the end gets the results . This procedure will takes a lot
of time.
algorithms are considered to be utilized in a good
range of fields including network login recognition, To save time and money which is needed for the initial
customer acquisition behaviour detection, diagnostic process, this program suggests an automatic
production process improvement, Mastercard fraud diagnostic system based on user input. The program takes
input from the user and provides an inventory of possible
detection, and disease predictions. Many of those diseases. In Fig.1 above we defined our proposed model.
applications are built using the ML supervised
The purpose of this study was to test the proposed hypothesis
method. During this way, records unit with labels that ML algorithms are monitored to improve health care with
had been delivered to hypothesis fashions that accurate and rapid diagnosis. In this study, we investigate
expect non-label fashions. This implies the thought studies using more than one monitored ML model for each
that, Medical doctors can use supervised learning as diagnostic problem. This way Provides more understanding
and accuracy because the performance testing of a single

978-1-6654-9602-5/22/$31.00 ©2022 IEEE 42

Authorized licensed use limited to: Florida Institute of Technology. Downloaded on February 05,2024 at 07:44:25 UTC from IEEE Xplore. Restrictions apply.
algorithm over different research settings creates biases that L1 and L2 , both have same length.
produce vague results. Analysis of ML models will be
performed on a few diseases of the heart, kidneys, breast, and
brain. To diagnose the disease, several methods are going to L1 S1 S2 S3 S4 S5 S6 S7 S8
be tested like KNN, NB, DT, CNN, SVM, and LR. We’ve
L2 0 0 0 0 0 0 0 0
used multinomial NB as many variants mean that more signals
are taken.
A. Machine Learning ii)Algorithm for Dataset part
The ML is a sub-field of AI. Its purpose is to enable Performing the same steps for both testing and training dataset
computers to read on their own.
Step 1: Using pandas module read the CSV file
The ML algorithm enables it to identify patterns in targeted
Step 2: Replace with index
data, create descriptive models and predict objects without
having clear predefined rules and models. S1 S2 S3 S4 S5 S6 S7 S8 … Prognosis
0 0 0 1 0 1 0 0 … 0
B. Different types of Machine Learning 0 1 0 0 0 0 0 0 … 1
i. Supervised Learning: -In supervised learning, we are
given a group of information and already know what 1 0 0 0 0 0 0 1 … 2
the proper output should appear as if, with the concept 0 0 0 0 0 0 0 0 … 3
that there’s a connection between the input and
output.
ii. unsupervised Learning: -It allows us to faced Step 3: Create X as symptoms and Y as disease.
problems with little or no knowledge of what our
results appears as if. We are able to experience the A. Naïve Bayes Algorithm
formation of knowledge after we don’t really know This program receives information from the user
the impact of the variables. We can find this structure and predicts the most likely diseases. This is
by combining data supported relationships between
variables within the data. achieved using datasets and machine learning
algorithms. The algorithm we use is Naïve
iii. Reinforcement learning: A range of software and
equipment are wont to determine the simplest Bayesian that works deterministically or
behaviour or approach. probabilistically. We need to import a scikit to read
the library to use. In this case, we use polynomial
III. IMPLEMENTATION
NB because there are many disease symptoms
We applied our algorithm to IDE Python 3.6.0 using the
VS code (visual studio).We tested the algorithm on a system Bayes theorem
with a quad core i5 processor , RAM-8gb, 1TB hard disk. • The objective of Bayes theory is to predict the
To analyse the result statistics, we used Matplotlib and disease category as a class label in our project with
therefore the NumPy python library. data in Tuple.
Symptoms are reported as infections. The disease is predicted
• Let X contain input H some assumptions, because
using Naive bayes algorithm. According to literature searches, tuple x (input (symptoms)) belongs to particular
this algorithm leads to high accuracy in large databases. The class c (output (disease)).
program contains labels and other illness has its own • In the case of separation problems, we check for
symptoms. 70% of the database is used for training and 30% is
used for testing. Training and testing is done on the system and the possibility that the dataset X is of type C, since
the results obtained are also available. we know that the attribute definition of X.
i)Algorithm for Initial part B. Dataset
Step 1: Import all the required packages i.e., Tkinter for GUI, The data comes from a study The study was
pandas for reading the csv files and numpy to perform carried out at the University of Colombia. It
numerical operations.
contains 150 diseases and each disease has an
Step 2: Creating a list which contains all the symptoms which average of 810 symptoms. 70% of the data used for
are according the dataset csv file
training is done taking into account all components
Step 3: Create another list which contains the diseases. of the input. The symptoms corresponding to the
Step 4: Create two empty list respective disease are marked as 1 and remain as 0.

43
Authorized licensed use limited to: Florida Institute of Technology. Downloaded on February 05,2024 at 07:44:25 UTC from IEEE Xplore. Restrictions apply.
Contains 5 drop down options where we have a list symptoms as a input from the user and provides accurately
predicted diseases as a output in within the text box.
box
of all successful symptoms. User can select one of
the symptoms and by clicking the button the The system will predict the disease in within which symptoms
correctly predicted disease will be displayed in the are given as an infection or anything adverse happening to you
which makes you uncomfortable.. The disease will be predicted
text box. Fig.2 contains a little detail of the dataset. using the Naive ve Bayesian algorithm. In keeping with a
literature study, this algorithm results in high accuracy of
enormous datasets. The GUI contains symptoms of all possible
diseases as labels, symptoms are chosen accordingly and then
prediction will be don.. 70% of the data
dataset will be used as
training and 30% will be used for data testing. Training and
testing will be done on the GUI and therefore the results
obtained are available.
• Diseases are predicted using the Naïve Bayesian
algorithm which works mainly with Multinomial NB
as many symptoms will be there.
there
• According to
o the literature, this algorithm leads to the
Fig.2. Datasets
complete accuracy of an outsized database.
• The GUI contains diseases like labels gives symptoms
C. GUI for each disease.
Tkinter GUI has been used in our project
project. It used for V. CONCLUSION
creating the GUI for desktop based applications.
The project is built so far that the system takes symptoms
Tkinter is very easy to grasp, required very little code for from the user as input and generates output i.e. disease
application, portable across all OS, and preinstalled with prediction. The user can select at least one to five symptoms.
Python library. And it is faster way to create GUI when python Less accuracy will be obtained if only one input is selected.
with Tkinter. The developed GUI is shown in Fig.3. The greater the number of input, the greater the accuracy. In
this paper we have proposed a learning model for a compact
novel machine Algorithm of Naïve bayes. We also tried to
reduce the number of features from the dataset. In this process
we were able to obtain sufficient
ient accuracy for all data sets
using our machine learning model. We found the best
accuracy of the most disease approx.(78.6%).

In upcoming Work, the event of complex ML algorithms is

extremely necessary for enhancement of disease prediction. In
additionally,
nally, data sets should be expanded to different
demographics to avoid overcrowding and to extend the
accuracy of the models used. Finally we’ll try and put all the
medical report into this specially last 10-20
10 medical records in
order that everything should
uld be smooth and our system will
curtain the burden of medical staff and lots of aspect.

ACKNOWLEDGMENT
Fig.3. Developed GUI
It’s an honor to work with Professor SPS Chauhan sir
(Program Chair), he guided us in all the obstacles, thankful for
IV. RESULT AND DISCUSSION the good support regarding paper and their all views. We are
Classical process could be a procedure within which a really thankful to our reviewer who encourages us to figure
patient should visit a doctor, receive various treatments and are more on this and showing their keen interest in our paper. I
at the end gets the results . This procedure will takes a lot of also place on record, my sense of gratitude to one and all, who
time. directly and indirectly, have lent their hand
h during this journey.
.
To save time and money which is needed for tthe initial
diagnostic process, this program suggests an automat
automatic REFERENCES
diagnostic system based on user input. The program takes

Authorized licensed use limited to: Florida Institute of Technology. Downloaded on February 05,2024 at 07:44:25 UTC from IEEE Xplore. Restrictions apply.
[1] M. Marimuthu, M. Abinaya, K. S., K. Madhankumar, and V. Pavithra, [7] P. P. Sengar, M. J. Gaikwad, and A. S. Nagdive, “Comparative study
“A Review on cardiovascular disease Prediction using Machine of machine learning algorithms for breast cancer prediction,”
Learning and Data Analytics Approach,” International Journal of Proceedings of the 3rd International Conference on Smart Systems and
Computer Applications, vol. 181, no. 18, pp. 20–25, 2018. Inventive Technology, ICSSIT 2020, pp. 796–801, 2020.
[2] Joshi J, Rinal D, Patel J, Diagnosis And Prognosis of Breast Cancer [8] Chen, Yixue Hao, Kai Hwang , Lu Wang, and Lin Wang(2017).
Using Classification Rules, International Journal of Engineering “Disease Prediction by Machine Learning Over Big Data From
Research and General Science,2(6):315-323, October 2014. Healthcare Communities”.IEEE Access. Vol. 5, pp. 8869-8879.
[3] V.Chaurasia and S. Pal, “Data Mining Approach to Detect Heart [9] Michael L. Raymer, William F. Punch, Erik D. Goodman, Leslie A.
Diseases”, International Journal of Advanced applied Science and Kuhn, and Anil K. Jain(2000). “Dimensionality Reduction Using
Information Technology (IJACSIT), Vol. 2, No. 4, 2013, Page 56-66. Genetic Algorithms”. IEEE Transactions on Evolutionary Computation.
[4] K. Dwivedi, “Performance evaluation of various machine learning Vol. 4, Issue 2, pp. 164-171.
techniques for prediction of heart condition,” Neural Computing and [10] D. Asir Antony Gnana Singh,E. Jebamalar Leavline R. Priyanka and P.
Applications, vol. 29, no. 10, pp. 685– 693, 2018. Padma Priya(2016).“Dimensionality Reduction using Genetic
[5] SolankiA.V., Data Mining Techniques using WEKA Classification for Algorithm for Improving Accuracy in Medical Diagnosis”, I.J.
Sickle Cell Disease, International Journal of Computer Science and Intelligent Systems and Applications MECS. No. 1, pp.67-73.
Information Technology,5(4): 58575860,2014. [11] S. Jadhav, R. Kasar, N. Lade, M. Patil, and S. Kolte, “Disease
[6] Ahmed F. Otoom, Emad E. Abdallah, Yousef Kilani, Ahmed Kefaye Prediction by Machine Learning from Healthcare Communities,”
and Mohammad Ashour.(2015) .Effective Diagnosis and Monitoring of International Journal of Scientific Research in Science and Technology,
Heart Disease. International Journal of Software Engineering and Its pp. 29–35, 2019.
Applications. Vol. 9, No. 1, pp. 143-156.

45
Authorized licensed use limited to: Florida Institute of Technology. Downloaded on February 05,2024 at 07:44:25 UTC from IEEE Xplore. Restrictions apply.