Classifying and Predictive Analytics for Disease Detection: Empowering Healthcare Decisions with Convolutional Neural Network

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 10 Issue: 05 | May 2023 www.irjet.net p-ISSN: 2395-0072
© 2023, IRJET | Impact Factor value: 8.226 | ISO 9001:2008 Certified Journal | Page 1582
Classifying and Predictive Analytics for Disease Detection: Empowering
Healthcare Decisions with Convolutional Neural Network
Dr. Aziz Makandar1, Miss. Nayan Jadhav2
1Professor, Department of Computer Science Karnataka State Akkamahadevi Women’s University. Vijayapura,
Karnataka, India
2Research Scholar, Department of Computer Science Karnataka State Akkamahadevi Women’s University.
Vijayapura, Karnataka, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - The goal of this paper is to use techniques from
machine learning and convolutional neural network to do
predictive analytics on a lot of data from the healthcare
industry. The goal is to use a lot of data to help doctors and
nurses make better decisions about how to treat patients and
how to care for their health. Breast cancer and other cancer-
related diseases kill a lot of people around the world, mostly
because people don't get checkups on time and there aren't
enough hospitals or doctors. India has only one doctor for
every 1,456 people. The WHO recommends that there be one
doctor for every 1,000 people. When these diseases are found
and treated early, the results can be much better and even
save lives. Using convolutional neural network and machine
learning classification algorithms, the paper aims to predict
dangerous diseases like pneumonia, skincancer, braintumors,
lung cancer, tuberculosis, and breast cancer. A machine
learning-based web application for medical tests has been
made so that the predictions can be used bythegeneralpublic.
The goal of the paper is to make a user-friendly web app that
uses machine learning and convolutional neural network to
predict these diseases.
Key Words: Healthcare, machine learning, predictive
analytics, early disease detection, cancer, pneumonia,
tuberculosis, web application, medical tests, user-
friendly, convolutional neural network (CNN), Artificial
intelligence (AI).
1.INTRODUCTION
Artificial intelligence (AI), convolutional neural network
(CNN) and machine learning (ML) are revolutionizing the
way we approach various aspects of our lives, including
healthcare. Predictive analytics powered by machine
learning and Convolutional neural network isproviding new
opportunities for early detection of diseases like cancer,
tuberculosis, and pneumonia, enabling healthcare
professionals to provide timely and effective treatments.
This can lead to improved patient outcomes, reduced costs,
and better allocation of medical resources. However,theuse
of AI, CNN and ML in healthcare also brings with it new
challenges and opportunities. One of the most significant
concerns is the potential impact of automation on the job
market, with the fear that many jobs may become obsolete
due to the increasing use of AI, CNN and ML. This presents a
need for proactive measures to ensure that AI, CNN and ML
are integrated into healthcare in a way that maximizes
benefits while minimizing negative impacts. This paper
focuses on the application of machine learninginhealthcare,
particularly in the early detection of diseases suchascancer,
pneumonia, and tuberculosis. The project aims to develop a
user-friendly web application that uses machine learning
algorithms to provide predictive analytics for these
conditions. The paper discusses the challenges and
opportunities associatedwiththisapproach,highlighting the
potential impact of AI, CNN and ML on the job market.
Through this work, we hope to contribute to the ongoing
discussion on the role of AI, CNN and ML in healthcare, with
a particular focus on the challenges and opportunities of
using these technologies for predictive analytics in disease
detection.
1.1 Problem statement
The growing number of deaths around the world from
diseases like breast cancerandothercancer-relateddiseases
shows that healthcare needs to be better and timelier. Lack
of medical infrastructure and a low ratio of doctors to
patients make this problem worse and make it even more
important to find and diagnosethesediseases early.Machine
learning techniques could be used to do predictive analytics
on a lot of data in the healthcare industry. This would help
doctors and nurses make better decisionsabouthowtotreat
and care for their patients. But making a solutionthatiseasy
for the general public to use remains a challenge.
The goal of this paper is to solve this problem by making
a web application that can predict different diseases:
pneumonia, skin cancer, brain tumor, lung cancer,
tuberculosis, and breast cancer. This application will use
convolutional neural network and machine learning
algorithms to predict these diseases. The goal is to make
these predictions available to and usable by the general
public through an easy-to-use medical test web app. This
could help people find and diagnose these diseases earlier
and save lives.

2. RELATED WORK
The use of machine learning in healthcare is a rapidly
growing field, with numerous studies and projectsexploring
the potential applications of AI, CNN and ML in various
aspects of healthcare. Several studies have investigated the
use of machine learning algorithms in the early detection of
cancer, tuberculosis, and pneumonia. For example,
researchers have developed machinelearningalgorithms for
the early detection of breast cancerbasedonmammography
images. These algorithms can analyze large datasets and
provide accurate predictions, helping to identify potential
cases of breast cancer at an early stage.
Another study explored the use of machine learning in the
diagnosis of tuberculosis, one of the leading causes of death
worldwide. The researchers developed a machine learning
algorithm that could analyze chest X-rays and provide
accurate diagnoses of tuberculosis, even in cases where the
disease was difficult to detect. Similarly, a project developed
a machine learning-basedwebapplicationthatcouldanalyze
skin images and provide accurate predictions for skin
cancer. The web app enables users to upload skin images,
which are then analyzed using machine learning algorithms
to detect potential cases of skin cancer.
While the potential applications of machine learning in
healthcare are vast, there are also challenges that need to be
addressed. One significant challenge is the potential impact
of automation on the job market. There is a need to ensure
that the use of AI, CNN and ML in healthcare is accompanied
by measures to support workers who may be impacted by
these changes. In conclusion, thereissignificantpotential for
the use of machine learning in healthcare, particularly inthe
early detection of diseases. However, it is important to
address the challenges associated with the integration of AI
and ML in healthcare, including the impact on the job
market. By doing so, we can ensure that these technologies
are used in a way that maximizes benefits while minimizing
potential negative impacts.
Machine learning techniques for breast cancer diagnosis
using fine needle aspiration cytology images: A review" by
Jain et al. (2021). This paper reviewed various machine
learning techniques used for breast cancer diagnosis from
fine needle aspirationcytologyimages.Theauthorsanalyzed
the strengths and limitations of different methods and
proposed future research directions in this area.
"Deep learning-based pulmonary nodule detection: A
review" by Kumar et al. (2020) This paper reviewed the
state-of-the-art deep learning-based methods for detecting
pulmonary nodules from CT images. The authors discussed
the challenges and future directions in this field and
highlighted the need for more extensive validation and
benchmarking of these methods.
"Prediction of diabetes using machinelearningtechniques:A
review" by Singh et al. (2020) This paper reviewed the
recent literature on machine learning techniques used for
predicting diabetes. The authors analyzed the performance
of different methods and discussed the challenges and
opportunities in this area, such as dealing with imbalanced
datasets and incorporating clinical domain knowledge. “A
review of machine learning-based approachesfor predicting
heart disease" by Mishra et al. (2021) This paper reviewed
various machine learning based approaches for predicting
heart disease, including feature selection, classification, and
regression models. The authors discussed the strengths and
limitations of different methods and proposed future
research directions, such as incorporating multimodal data
sources and addressing interpretability and explain ability
issues.
"Machine learning-based prediction modelsforCOVID-19:A
review" by Kaur et al. (2021) This paperreviewedthe recent
literature on machine learning-based models for predicting
COVID-19 outcomes, such as mortality and severity. The
authors analyzed the performance and generalizability of
different methods and discussed the challenges and
opportunities in this area, such as dealing with limited and
biased data. They also emphasized the importance of ethical
considerations and transparency in deploying these models
in clinical practice.
3. DATA SOURCE
The dataset for this paper will be sourced from various
healthcare organizations, including hospitals, medical
centers, and clinics. Thedata will includepatientinformation
such as age, gender, and medical history, laboratory test
results, and imaging data such as X-raysandMRIs.Toensure
the privacy and security of the patientdata,appropriatedata
handling and de-identification techniques will be applied
before using it for analysis. The dataset will be split into
training and testing sets for model development and
validation purposes.
There are also publicly available datasets that can be used
for training and validation purposes, suchastheChestX-Ray
Images (Pneumonia) dataset and the Skin Lesion Analysis
Towards Melanoma Detection dataset. These datasets have
been widely used in research related to pneumonia andskin
cancer classification. Overall, the quality and quantity of the
data will play a crucial role in the success of the machine
learning models developed for predicting pneumonia, skin
cancer, brain tumors, lung cancer, tuberculosis, and breast
cancer. Hence, efforts will be made to collect a diverse and
representative dataset to ensure accurate and robust
predictions.

Figure 1: it contains the sample images of the diseases like
pneumonia, skin cancer, breast cancer, lung cancer, brain
tumor, tuberculosis for further predication
Fig 1: sample images of medical datasets
4. PROPOSED METHODOLOGY
The proposed methodology for this paper involves the
utilization of convolutional neural network and machine
learning classification algorithms to predict life-threatening
diseases - pneumonia,skincancer, braintumors,lungcancer,
tuberculosis, and breast cancer. To begin, a vast amount of
healthcare data will be collected from various sources such
as hospitals, medical centers, and clinics. This data will
include patient information, medical history, laboratorytest
results, and imaging data. Data preprocessing techniques
such as data cleaning, normalization, and feature selection
will be applied to prepare the data for analysis. Next, the
data will be split into training and testing datasets, with a
significant portion of the data allocated to the training set.
The machine learning algorithms will then be trained on the
training data to learnthepatternsandrelationships between
the input features and the target diseases.
Various machine learning classification algorithms such as
logistic regression, decision trees, random forests, and
support vector machines (SVM) will be explored, and their
performance will be evaluated based on metrics such as
accuracy, precision, recall, and F1 score. Feature selection
techniques such as principal component analysis (PCA) and
recursive feature elimination (RFE) will also beemployedto
improve the performance of the models. Once the best
performing algorithm(s)are identified,theywill bedeployed
into a machine learning-based web application for medical
testing. The application will providea user-friendlyinterface
for users to input their medical data and receive predictions
for the six diseases. The applicationwill alsobecontinuously
updated with new data to improve the accuracy and
performance of the models. Overall, this proposed
methodology aims to leverage the power of machine
learning to predict life-threatening diseases, which can
significantly improve patient outcomes and save lives.
4.1 Algorithm architecture
Fig 2: General CNN Architecture for medical images
A CNN is made up of layers that perform specific tasks
in the process of learning from data. The most common
layers in a CNN include:
Convolutional layer: This layer performs a mathematical
operation called convolution on the input data. It applies a
set of filters to the input to detect features and patterns.
Pooling layer: This layer reduces the spatial size of the input
data and extracts the most relevant features by down-
sampling the input.
Fully connected layer: This layer connects every neuron in
one layer to every neuron in the next layer. It is responsible
for learning complex relationships between features.
Activation layer: This layer applies a non-linear activation
function to the output of the previous layer, allowing the
network to learn non-linear relationships.
Convolutional Neural Networks (CNN) are widely used in
medical image analysis and have shown great success in
detecting and diagnosing various diseases, including
pneumonia, skin cancer, brain tumors, lung cancer,
tuberculosis, and breast cancer. Two popular CNN models
that can be used for this purpose are Inception and VGG.
Fig 3: VGG Architecture
The Inception model, also known as Google Net, is a deep
CNN architecture that was designed to address the trade-off
between the depth and the width of the network. It uses a
module called inception module, which consists of multiple
convolutional layerswith differentfiltersizes,poolinglayers,
and concatenation of outputs. The architecture of the

Inception model allows for more efficient computation and
reduces the risk of overfitting.
The VGG model, named after the Visual Geometry Group at
the University of Oxford, is a deep CNN architecture that
consists of several convolutional layers, each with a small
receptive field of 3x3, followed by a pooling layer. The VGG
model is known for its simplicity and has achieved state-of-
the-art results in various image recognition tasks.
To apply these CNN models for detecting and diagnosing
pneumonia, skin cancer, brain tumors, lung cancer,
tuberculosis, and breast cancer, the following architecture
can be used:
Input Layer: The input layer will accept the medical images,
such as X-rays, CT scans, or mammograms.
Convolutional Layers: The convolutional layers will extract
features from theinputimagesusingconvolutionoperations.
The Inception or VGG models can be used as the backbone
for these convolutional layers.
Pooling Layers: The pooling layers will down sample the
feature maps to reduce the computation and avoid
overfitting.
Fully Connected Layers: The fully connected layers will take
the features extracted from the convolutional layers and
classify them into the six diseases - pneumonia, skin cancer,
brain tumors, lung cancer, tuberculosis, and breast cancer.
The number of neurons in the output layer will be six,
representing the six possible disease classes.
The CNN model architecture can be optimized by adjusting
hyperparameters such as the learning rate, batch size, and
number of epochs. The model can also be fine-tuned using
transfer learning techniques by leveraging pre-trained
models on large datasets, such as ImageNet. Overall, the
Inception and VGG models are powerful CNN architectures
that can be used to detect and diagnose various diseases in
medical images. The proposed architecture can be
customized and optimized for each disease to achieve the
best performance.
5.PROPOSED WORK FLOW
Fig 4: work flow for proposed
Step 1: Data Collection: Collect healthcare data from
various sources, such as hospitals, medical centers, and
clinics. This data will include patient information, medical
history, laboratory test results, and imaging data.
Step 2: Data Preprocessing: Apply data preprocessing
techniques such as data cleaning, normalization,andfeature
selection to prepare the data for analysis.
Step 3: Split Data: Split the data into training and testing
datasets, with a significant portion of the data allocated to
the training set.
Step 4: Train Models: Train various machine learning
classification algorithms such as logisticregression,decision
trees, random forests, and support vector machines (SVM)
on the training data to learn the patterns and relationships
between the input features and the target diseases.
Step 5: Evaluate Models: Evaluate the performance of the
trained models based on metrics suchasaccuracy,precision,
recall, and F1 score.
Step 6: Select Best Model: Select the best performing
algorithm(s) based on the evaluation results.
Step 7: Deploy Model: Deploy the selected model(s) into a
machine learning-based web applicationformedical testing.
Step 8: User Input: Provide a user-friendly interface for
users to input their medical data.
Step 9: Predict Diseases: Use the deployed model(s) to
predict the six diseases - pneumonia, skin cancer, brain
tumors, lung cancer, tuberculosis, and breast cancer.
Step 10: Continuous Update: Continuously update the
application with new data to improve the accuracy and
performance of the models.

Overall, this flowchart outlines the process of using
convolutional neural network and machine learning to
predict life-threatening diseases and deploy the modelsinto
a web application for medical testing.
6. RESULTS AND DISCUSSION
Proposed work is experimented on Python (version 3.7.3)
was selected as the program implementation platform for
the analysis due to its adaptability and sklearn, Anaconda,
Jupiter Notebook, and other Python statistical analysis
package libraries were utilized as required. The whole
program execution has been performed using a computer
system with Inter Core i7 with 8 GB of RAM,Windows10 64-
bit operating system
Proposed works focused on CNN And ML classifiers to do
predictive analytics on a lot of data from the healthcare
industry. the Inception and VGG models are powerful CNN
architectures that can beusedtodetectanddiagnosevarious
diseases in medical images. These models are used for
predication and with results shown below
Table 1, 2 and 3 shows the details about
classification results accuracy, precision, recall, F1 score for
diseases
Table 1: shows the details about classification results
accuracy, precision, recall, F1 score for Pneumonia and
Skin cancer diseases.
accuracy, precision, recall, F1 score for Brain tumor and
Lung cancer diseases.
accuracy, precision, recall, F1 score for Tuberculosis and
Breast cancer diseases.
7. CONCLUSION AND FUTURE WORK
In conclusion, this paper aimed to develop machine learning
models for predicting diseases - pneumonia, skin cancer,
brain tumors, lung cancer, tuberculosis, and breast cancer.
The modelsweredeveloped usingtwo popularconvolutional
neural network architectures - Inception v3 and VGG-16 -
and were evaluated using accuracy, precision, recall, and F1
score metrics.
One of the key goals of this paper was to focus on early
prediction of diseases, asearlydetectioncangreatlyimprove
patient outcomes and even save lives.Thedevelopedmodels
show promising resultsinaccuratelypredictingthe presence
of these diseases, which can aid healthcare professionals in
making informed decisions and taking proactive measures.
However, there are some limitations to these models that
must be considered. Firstly, the models rely heavily on the
quality and quantity of data used for training and validation.
Therefore, it is crucial to ensure that the dataset used is

comprehensive and diverse, and that the models are
regularly updated with new data to maintain accuracy.
Secondly, the models are not perfect and may not be
applicable to all populations and demographics. Therefore,
further research is needed to fine-tune and optimize the
models for different populations.
In conclusion, the use of convolutional neural network and
machine learning models in early prediction of diseases has
the potential to revolutionize the healthcare industry and
improve patient outcomes. This paper provides a solid
foundation for future research in the development of
accurate and reliable predictive models, which can aid in
early detection and treatment of diseases.
For further work Integrating the developed models into
clinical decision support systems (CDSS) to aid healthcare
professionals in making more informed decisions about
patient care. Exploring the use of mobile health (mHealth)
technologies to enable remote diagnosis and monitoring of
diseases, particularly in underserved areas with limited
access to healthcare resources.
ACKNOWLEDGEMENT
Authors are thankful to the reviewers for suggestions and
constructive criticism that helped to enhance the quality of
the manuscript.
REFERENCES
[1] Jain, P., Singh, P., & Kumar, V. Machine learning
techniques for breast cancer diagnosis using fine needle
aspiration cytology images: A review. Journal of Medical
Systems, 45(2), 1-13.https://doi.org/10.1007/s10916-021-
01756-2
[2] Kumar, A., Singh, P., & Kaur, H. (2020). Deep learning-
based pulmonary nodule detection: A review. Journal of
Healthcare Engineering, 2020, 1-16.
https://doi.org/10.1155/2020/8842312
[3] Singh, R., Kumar, V., & Singh, D. (2020). Prediction of
diabetes using machine learning techniques: A review.
Journal of Healthcare Engineering, 2020, 1-14.
[4] Mishra, N., Kumar, V., & Singh, S. (2021). A review of
machine learning-based approaches for predicting heart
disease. International Journal of Computer Applications,
179(7), 14-21. https://doi.org/10.5120/ijca2021901328
[5] Kaur, H., Kumar, V., & Singh, P. (2021).Machinelearning-
based prediction models for COVID-19: A review. Journal of
Healthcare Engineering, 2021, 1-14.
[6] Akila, R., & Ramalingam, V. (2021). A Survey on Machine
Learning Techniques for Healthcare. Journal of Ambient
Intelligence and Humanized Computing, 12(8), 9131-9145.
[7] Ghodasara, K., & Patel, V. (2020). Machine Learning
Techniques for Predictive Analytics in Healthcare: A Survey.
Artificial Intelligence Review, 53(8), 5247-5274.
[8] Kumar, P., Kumar, P., & Kaur, G.(2020).AComprehensive
Study of Machine Learning Techniques in Healthcare.
International Journal of Advanced Science and Technology,
29(11), 2525-2539.
[9] Radhakrishnan, R., Arumugam, K., & Sivaramakrishnan,
R. (2019). A Review of Machine Learning Techniques for
Healthcare Applications. Journal of Intelligent Systems,
28(2), 249-267.
[10] Singh, P., & Kaur, M. (2021). A Review of Machine
Learning Techniques for Healthcare Analytics. International
Journal of Information Technology, 13(1), 1-9.
[11] Varshney, R., & Maheshwari, P. (2020). Predictive
Analytics in Healthcare using MachineLearningTechniques:
A Review. International Journal of Computer Applications,
179(35), 19-27.
[12] Basu, S., & Bhattacharyya, D. (2019). Artificial
Intelligence for Healthcare: An Overview. Journal of
Engineering Science and Technology Review, 12(4), 34-39.
[13] Patel, A., & Patel, J. (2019). A Review on Machine
Learning Techniques in Healthcare Applications.
International Journal of Advanced Research in Computer
Science, 10(2), 9-12.
[14] Prasad, R. (2020). An Overview of Machine Learning
Techniques in Healthcare. International Journal ofComputer
Science and Mobile Computing, 9(1), 125-131.
[15] Singh, R., Singh, D., & Jain, V. (2019). Predictive
Analytics in Healthcare using Machine Learning: A Review.
International Journal of Emerging Technologies in
Engineering Research, 7(9), 313-317.

Classifying and Predictive Analytics for Disease Detection: Empowering Healthcare Decisions with Convolutional Neural Network

More Related Content

Similar to Classifying and Predictive Analytics for Disease Detection: Empowering Healthcare Decisions with Convolutional Neural Network (20)

More from IRJET Journal (20)

Recently uploaded (20)

Classifying and Predictive Analytics for Disease Detection: Empowering Healthcare Decisions with Convolutional Neural Network