© 2023 IJNRD | Volume 8, Issue 4 April 2023 | ISSN: 2456-4184 | IJNRD.

ORG

MULTIPLE DISEASE PREDICTION SYSTEM

USING MACHINE LEARNING
1 Prof. DR. R. Srinivasa Rao, 2 E. Bhargavi, 3 D. Purnachandra Rao, 4 B. Kanthi Kumar, 5 D. Venkatesh

1 Professor, Department of ECE, SRGEC, Gudlavalleru,

2 Undergraduate Student, Department of ECE, SRGEC, Gudlavalleru,

3 Undergraduate Student, Department of ECE, SRGEC, Gudlavalleru,
4 Undergraduate Student, Department of ECE, SRGEC, Gudlavalleru,
5 Undergraduate Student, Department of ECE, SRGEC, Gudlavalleru,

Abstract:
People are becoming increasingly susceptible to high-risk illnesses like chronic diabetes, heart disease, Parkinson’s
disease etc. The mortality ratio is prevalent nowadays due to the enormous number of deaths. Many of the current
machine learning models for health care analysis focus on a single disease at a time. One analysis is for diabetes, one
for the heart, and one for Parkinson's of that nature. There is no standard system that allows one analyst to forecast
more than one disease at a time. They also provide a result with low accuracy and precision. Then that lower accuracy
puts the patient's life in danger.
Hence, using machine learning we are suggesting a predictive system, the so-called Multiple Disease Prediction
System which is used to predict diseases accurately and simultaneously forecast several diseases in one shot. Diabetes,
heart disease, and Parkinson's disease are the three diseases we have currently taken into consideration. Future, many
more diseases may be added. Here, we employ machine learning (ML) techniques like SVM for (diabetes and
Parkinson's) and Logistic Regression for Heart Disease for diseases prediction. The user must enter many disease-
related parameters before the system displays a result indicating whether the user has the disease or not.
It aids in the processing of the massive volumes of data produced by the medical sector and significantly shortens the
time it takes for doctors to detect a patient's disease at an early stage. It also encourages the adoption of preventative
measures to lengthen the patient's life expectancy.
Key Phrases: Machine Learning, Artificial Intelligence, SVM, Logistic Regression, Diabetes, Heart, Parkinson’s.
1. INTRODUCTION
With the advent of the digital era and new technical breakthroughs, necessary data is computerizing for rapid access
and helps to transport files to distant distances within seconds. In this digital world, enormous amounts of data are
generated day by day and become an asset. So, the medical sector is one of those industries that generates enormous
amounts of data. The patient's data about their symptoms, along with clinical factors, hospital resources, diagnostic
data, patient’s records, and medical equipment, make up the created data in the health care industry. To extract
knowledge for precise decision-making, huge, dense, and complicated data must be processed and assessed. With the
IJNRD2304131 International Journal of Novel Research and Development (www.ijnrd.org) b270
 © 2023 IJNRD | Volume 8, Issue 4 April 2023 | ISSN: 2456-4184 | IJNRD.ORG
help of different algorithms like Decision Tree, Random Forest, SVM, Logistic Regression, and others, medical data
mining has a great deal of potential for revealing hidden patterns, correlations, and relationships between features in
medical data sets to instantly predict the disease from the patient's symptoms.
Several of the current models focus on just one disease for each analysis. One analysis, for instance, may cover
diabetes, one analysis may cover heart, and another analysis for Parkinson's disease. There would be no shared system
that could simultaneously analyze multiple diseases.
As a result, we are producing Multiple Disease Prediction System which predicts more than one disease at a time.
Here, we are considering giving consumers immediate access to precise disease predictions based on the symptoms
they enter. As a result, we are presenting a method that makes use of Streamlit to forecast various diseases. We will
examine diabetes, heart disease, and Parkinson's disease analyses in this system. Later, many more illnesses could be
added. We will use machine learning algorithms, pickle module, streamlit to implement a multiple disease prediction
system. The Python pickling library is used along with the algorithms SVM and Logistic Regression. Model behavior
is saved using the Pickling library. An open-source framework called Streamlit is used to create online applications
without any prior HTML, CSS, or JavaScript expertise.
The user must supply both the disease name and its parameters when he or she accesses this UI. Invoking the
appropriate model, Streamlit will then report the patient's state.
1.1 Description:
Many of the old technologies being used to study data in the medical field only analyze one disease at a time. One
system might be used to study diabetes, another to study Parkinson's, and yet another to forecast heart problems. Most
of them specialize in a different ailment. A hospital must use more models when it is to test the patient for more
disease at a time. A user can concentrate on more diseases using the same webpage using multiple diseases prediction
system. To determine whether the user is ill, with the proposed system, the user does not need to travel to several
locations. So, the user must choose their favorite disease in MDPS.
1.2 Problem system:
Most of the time, systems utilizing ML can only predict one disease at a time. If we need to test a patient immediately
to diagnose him or her for multiple diseases but cannot locate a system that can forecast more than one disease at a
time, we will need to use more models, which requires more time and money and has poorer accuracy, endangering
the patient's life with false results.
1.3 Proposed System:
Now-a-days Machine learning is rapidly advancing with a variety of algorithms that can process enormous amounts
of data, making it possible to automate tasks without the need for human intervention and deliver accurate results
right away. Hence, to study many diseases concurrently, we are creating a method called the Multiple Disease
Prediction System. The user does not have to visit several websites to research the diseases consequently. For the
time being, we have focused on Parkinson's illness, chronic diabetes, and hearing disorders. Later, other illnesses
may be added to the MDPS. We are employing ML techniques and Streamlit to implement MDPS. The user must
supply the disease parameter along with the disease when accessing the UI.
2. LITERATURE REVIEW
1. The report claims that one of the chronic disorders when blood sugar (glucose) levels are elevated is diabetes. It
causes a wide range of illnesses, such as blindness, among others. As it is simple and adaptable to predict if the patient
has the condition or not, they have employed ML approaches in the suggested work to analyze diabetes disease. The
system's introduction is primarily driven by the need to accurately diagnose diabetes in patients. SVM (Support
Vector Machine) and Logistic Regression are the two algorithms employed in the prediction system, and they have
accuracy ratings of 78%, 75% respectively. They compared two models' accuracy here [1].
2. The heart plays a crucial role in people, which is the motivation behind the suggested paper. Since heart-related
illnesses are on the rise nowadays, it is crucial that they are accurately diagnosed and predicted because they can
result in fatal heart complications. Hence, recent developments in AI and ML can enable developing a system that
accurately and quickly forecasts the disease. So, using datasets acquired from the well-known website Kaggle, the
IJNRD2304131 International Journal of Novel Research and Development (www.ijnrd.org) b271
 © 2023 IJNRD | Volume 8, Issue 4 April 2023 | ISSN: 2456-4184 | IJNRD.ORG
authors of this research analyze the accuracy of machine learning (ML) for predicting heart disease using logistic
regression, diabetes, and Parkinson's disease using SVM. They also contrasted the methods using SVM (81% and
Logistic Regression 82%) accuracy as a benchmark [2].
3. The system targets dopamine deficiency-related Parkinson's disease causes. Dopamine is a vital brain chemical
that is thought to help nerves communicate and is known as a neurotransmitter. Parkinson's disease has no known
cure, however there are ways to manage the symptoms. Hence, we can quickly identify Parkinson's illness utilizing
automated systems and machine learning techniques. For Parkinson's disease prediction, they employed SVM.
Accuracy is 87%[3].
3. SYSTEM ANALYSIS
3.1 Functional Requirement:
 The system ought to provide early disease detection by enabling patient disease prediction.
 The user must select the desired ailment from the available list of input values, and the output will be
displayed based on the trained model of the user input.
3.2 Non-Functional Requirement:
 A range of values during the prediction and diagnosis will be provided by the website.
 The website needs to be trustworthy and reputable.

4. DESIGN

4.1 Architecture Design:

Figure No 4.1: Block Diagram

Fig 4.1., shows architecture diagram for diabetes prediction model. This model has six different modules. These
modules include
1. Dataset Collection
2. Data Pre-processing

IJNRD2304131 International Journal of Novel Research and Development (www.ijnrd.org) b272

 © 2023 IJNRD | Volume 8, Issue 4 April 2023 | ISSN: 2456-4184 | IJNRD.ORG
3. Data Analysis
4. Train Test Split
5. Build Model
6. Evaluation
As diabetes, heart disease, and Parkinson's disease are all related to one another, we conducted experiments on these
three disorders in figure 4.1. The dataset for diabetes, heart disease, and Parkinson's disease must be accessed first.
The PIMA dataset, the heart disease dataset, and the Parkinson's dataset have all been imported from the Kaggle
Website which is the best source for the datasets. After the dataset has been imported, the individual input data is
visualized. After data pre-processing for visualization, where we look for outliers, missing values, and scale the
dataset, we divide the data into training and testing on the updated dataset.
The training dataset was used to apply SVM and Logistic algorithms, while the testing dataset was used to apply
knowledge of the classified algorithm. We will select the algorithm with the highest accuracy for each ailment after
applying our knowledge. After that, we create a pickle file for each disease and connect it with the Streamlit
framework to provide the model's output on the website.
4.2 User Interface Design:

Figure No 4.2: Graphical User Interface

5. IMPLEMENTATION
5.1 Algorithm

5.1.1. Logistic Regression Algorithm:

 Logistic Regression is a supervised learning model
 Classification model
 Uses sigmoid function
 Uses for Binary Classification problems (0 or 1, yes or no, high or low)
 Uses Binary Cross Entropy Loss Function (or) Log Loss Function

IJNRD2304131 International Journal of Novel Research and Development (www.ijnrd.org) b273

 © 2023 IJNRD | Volume 8, Issue 4 April 2023 | ISSN: 2456-4184 | IJNRD.ORG
Gradient Descent: Gradient Descent is an optimization algorithm used for monitoring the cost function in various
machine learning algorithms. It is used for updating the parameters of the learning model
Loss Function: Difference between the estimated value and true value is called Loss Function. Logistic Regression
uses Binary Cross Entropy Loss Function (or) Log Loss function.

Cost Function: Which deals with a penalty for a number of training sets or the complete batch.

Operation of Logistic Regression:

Step 1: Adjust the Logistic Regression Classifier to the preprocessed dataset.
Step 2: The classifier fits the dataset to create the prediction model. The best weight and bias parameters are provided
by the built model for precise prediction.
Step 3: The following procedures can be used to determine the ideal parameters
Equations for building models:
1. Sigmoid Function:

2. Updating weights through Gradient descent:

w2 = w1 – L*dw
b2 = b1 – L*db
3. Derivatives:
dw = 1/m(y_hat-y).X
db = 1/m(y_hat-y)
Step 4: The model has been trained and is now capable of disease prediction.
Step 5: By entering the user parameters and the anticipated disease, the patient's condition can be determined.
w: weight, b: bias, x: input feature, y hat: expected/predicted value (probability of y being 1), L: Learning Rate, Z:
Straight line equation, m: Total no of training examples, dw: Partial Derivative of cost function with respect to weight,
db: Partial Derivative of cost function with respect to Bias, y: True value, w2: Denotes a new weight, b2: A new bias
value, w1: A previous weight and b1: Bias.
Note: The appropriate weight and bias values can be found using the Gradient Descent Algorithm in order to
determine the minimum cost function.

IJNRD2304131 International Journal of Novel Research and Development (www.ijnrd.org) b274

 © 2023 IJNRD | Volume 8, Issue 4 April 2023 | ISSN: 2456-4184 | IJNRD.ORG
5.1.2. Support Vector Machine Algorithm:
 Model for Supervised Learning
 Classification and regression both
 Sufficient for High Dimensional Dataset
 Uses Hinge Loss Function
Hyperplane: A hyperplane is a line or plane that divides the data points into two classes in two-dimensional (2D)
space.
Support Vectors: The datapoints that are closest to the hyperplane are support vectors. The hyperplane's position
changes if these data points change.

Margin: The distance between two lines on distinct classes' nearest data points.

Maximum margin: An ideal hyperplane is one that has the largest margin.

Equation of Maximum Margin:

Loss Function : Loss function measures how far an estimated value is from its true value. It is helpful to determine
which performs better & which parameters are better.

Hinge Loss: Support Vector uses the Hinge Loss function which is one of the types of Loss function, mainly used
for maximum margin classification models. Hinge Loss incorporates a margin or distance from the classification
boundary into the loss calculation. Even if new observation are classified correctly, they can incur a penalty if the
margin from the decision boundary is not large enough

Equation of Hinge Loss Function: L = MAX(0,1 - yi(w.T*xi + b))

Operation of SVM:
Step 1: Fit the Support Vector Machine Classifier to the preprocessed dataset.
Step 2: The classifier fits the dataset to create the prediction model. The best weight and bias parameters are provided
by the built model for precise prediction.
Step 3: The following procedures can be used to determine the ideal parameters:
Equations for building models:
1. Equation of the Hyperplane:
y = w*x - b
2. Updating weights through Gradient descent:
w2 = w1 – L*dj/dw
b2 = b1 – L*dj/db
3. Derivatives:

IJNRD2304131 International Journal of Novel Research and Development (www.ijnrd.org) b275

 © 2023 IJNRD | Volume 8, Issue 4 April 2023 | ISSN: 2456-4184 | IJNRD.ORG
If( yi. (w*x + b) >= 1): else (yi.(w.x+b)<1):
dj/dw = 2*lambda*w dj/dw = 2*lambda*w – yi.xi
dj/db = 0 dj/db = yi
Step 4: The model has been trained and is now capable of disease prediction.
Step 5: By entering the user parameters and the anticipated disease, the patient's condition can be determined.
w: weight, b: bias, x: input feature, y hat: expected/predicted value, L: Learning Rate, dj/dw: Partial Derivative of
cost function with respect to weight, dj/db: Partial Derivative of cost function with respect to Bias, y: true value,
lambda: regularization parameter, w2: denotes a new weight, b2: a new bias value, and w1: a previous weight and
b1: old bias.
Note: The appropriate weight and bias values can be found using the Gradient Descent Algorithm in order to
determine the minimum cost function.
6. RESULT
Three diseases are included in the Multiple Disease Prediction System: diabetes, heart disease, and Parkinson's. SVM
is used to predict diabetes and Parkinson's disease, and logistic regression is used to predict heart disease for greater
accuracy. The system returns the patient's state once the user provides the relevant parameters and the projected
disease. The warning indication that adds a correct value will appear if the value fields are out of range or empty.
ACCURACY FOR EACH DISEASE:
Table No 6.1: Diabetes Disease

ALGORITHM DIABETES_DISEASE
Support Vector Machine 78%
Logistic Regression 75%

Table No 6.2: Heart disease

ALGORITHM HEART_DISEASE
Support Vector Machine 81%
Logistic Regression 82%

Table No 6.3: Parkinson’s disease

ALGORITHM PARKINSON’S_DISEASE
Support Vector Machine 87%
Logistic Regression 87%

IJNRD2304131 International Journal of Novel Research and Development (www.ijnrd.org) b276

 © 2023 IJNRD | Volume 8, Issue 4 April 2023 | ISSN: 2456-4184 | IJNRD.ORG

1. Diabetes Disease:

Figure No 6.1: The person is Diabetic

Figure No 6.2: The person is Not Diabetic

IJNRD2304131 International Journal of Novel Research and Development (www.ijnrd.org) b277

 © 2023 IJNRD | Volume 8, Issue 4 April 2023 | ISSN: 2456-4184 | IJNRD.ORG
2. Heart Disease:

Figure No 6.3: The person has heart disease

Figure No 6.4: The person does not have heart disease

IJNRD2304131 International Journal of Novel Research and Development (www.ijnrd.org) b278

 © 2023 IJNRD | Volume 8, Issue 4 April 2023 | ISSN: 2456-4184 | IJNRD.ORG
3. Parkinson’s Disease:

Figure No 6.5: The person has Parkinson’s disease

Figure No 6.6: The person does not have Parkinson’s disease

7. CONCLUSION
The goal of this research was to develop a multi-disease prediction system that would provide accurate predictions
right away. This project allows users to handle enormous amounts of data more quickly and without having to
navigate through numerous websites, both of which add to the speed of prediction. In addition to increasing human
life expectancy and preventing financial hardship, early disease prediction can help us take preventative actions,
which in turn lowers the mortality ratio. To achieve the highest level of accuracy when compared to other
classification algorithms like decision tree, Naive Bayes, Random Forest, KNN, we used machine learning algorithms
like Support Vector Machine and Logistic Regression for implementing Multiple Disease Prediction System.

IJNRD2304131 International Journal of Novel Research and Development (www.ijnrd.org) b279

 © 2023 IJNRD | Volume 8, Issue 4 April 2023 | ISSN: 2456-4184 | IJNRD.ORG

Figure No 7.1: Results of classification algorithms for predicting diabetes disease accuracy

Figure No 7.2: Results of classification algorithms for predicting heart disease accuracy

IJNRD2304131 International Journal of Novel Research and Development (www.ijnrd.org) b280

 © 2023 IJNRD | Volume 8, Issue 4 April 2023 | ISSN: 2456-4184 | IJNRD.ORG

Figure No 7.3: Results of classification algorithms for predicting parkinson’s disease accuracy

8. FUTURE SCOPE
 The current User Interface could eventually be expanded to include a considerable number of ailments.
 To extend life expectancy, we can work to boost prediction accuracy.
 Try to provide a dependable, approachable, and consistent user interface that includes illness stages and the
essential safety measures based on that disease degree.
ACKNOWLEDGEMENT
We would like to express our gratitude to Dr. G. V. S. N. R. V. Prasad, the principal of Seshadri Rao Gudlavalleru
Engineering College, for providing us with the opportunity and time to conduct and study on the topic of "Multiple
Disease Prediction System Using Machine Learning." We appreciate DR. R. Srinivasa Rao, who served as our
mentor, and Prof. Y. Rama Krishna, the department's head of electronics and communication, for their assistance
during our research, which would have seemed challenging without their inspiration, ongoing support, and insightful
suggestions. Without the cooperation, advice, and assistance of our friends and family, the complexity of this study
article would not have been achievable.
REFERENCES
[1] Laxmi Deepthi Gopisetti, Srinivas Karthik Lambavai Kummera, Sai Rohan Pattamsetti, Sneha Kuna, Niharika
Parsi, Hari Priya Kodali, “Multiple Disease Prediction Model by using Machine Learning and Streamlit” 2023 IEEE,
5th International Conference on Smart Systems and Inventive Technology (ICSSIT)
[2] Akkem Yaganteeswarudu, “Multi Disease Prediction Model by using Machine Learning” 2020 IEEE, 5th
International Conference on Communication and Electronics Systems (ICCES)

[3] Elsevier B.V,” Diabetes Prediction Using Machine Learning” 2019, International Conference on Recent Trends
in Advanced Computing.

[4] KM Jyoti Rani, “Diabetes Prediction Using Machine Learning” July 2020, International Journal of Scientific
Research in Computer Science Engineering, and Information Technology

IJNRD2304131 International Journal of Novel Research and Development (www.ijnrd.org) b281

 © 2023 IJNRD | Volume 8, Issue 4 April 2023 | ISSN: 2456-4184 | IJNRD.ORG
[5] Firdous, Shimoo, Wagai, Gowher A, Sharma, Kalpana, “A survey on diabetes risk prediction using machine
learning approaches”, November 2022, Journal of Family Medicine, and Primary Car.
[6] Krittanawong, C. Virk, H. U., Bengaluru, S., Wang, Z., Johnson, K. W., Pinotti, R., Zhang, H., Kaplin, S.,
Narasimhan, B., Kitai, T., Baber, U., Halperin, J. L., & Tang, W. H. (2020). Machine learning prediction in
cardiovascular diseases.
[7] Chaimaa Boukhatem, Heba Yahia Youssef, Ali Bou Nassif. February 2022 IEEE, Advances in Science and
Engineering Technology International Conferences (ASET)
[8] Supriya Kamoji, Dipali Koshti, Valiant Vincent Dmello, Alrich Agnel Kudel, Nash Rajesh Vaz, Prediction of
Parkinson's Disease using Machine Learning and Deep Transfer Learning from different Feature Sets, July 2021
IEEE, 6th International Conference on Communication and Electronics Systems (ICCES).
[9] Rohit Surya, A.T., Yaswanthram, P., Nair, P.R., Rajendra Prasath, S.S., Akella, S.V. (2022). Prediction of
Parkinson’s Disease Using Machine Learning Models—A Classifier Analysis. In: Bianchini, M., Piuri, V., Das, S.,
Shaw, R.N. (eds) Advanced Computing and Intelligent Technologies. Lecture Notes in Networks and Systems, vol
218. Springer, Singapore. https://doi.org/10.1007/978-981-16-2164-2_35.
[10] Makarious, M. B., Leonard, H. L., Vitale, D., Iwaki, H., Sargent, L., Dadu, A., Violich, I., Hutchins, E., Saffo,
D., Kim, J. J., Song, Y., Maleknia, M., Bookman, M., Nojopranoto, W., Campbell, R. H., Hashemi, S. H., Botia, J.
A., Carter, J. F., Craig, D. W., . . . Nalls, M. A. (2022). Multi-modality machine learning predicting Parkinson’s
disease. Npj Parkinson's Disease, 8(1), 1-13. https://doi.org/10.1038/s41531-022-00288-w.

IJNRD2304131 International Journal of Novel Research and Development (www.ijnrd.org) b282