Human Disease prediction using Machine Learning

Saksham Suman , Rahul Thakur , Saurabh Kumar , Rahul Kumar Sharma , Abhishek Gupta
Dept.of Electronics and Communications Engineering Assam University Silchar Silchar(India)

Abstract—Improving the care and treatment results of patients records, imaging, and genetic information. One major benefit
requires early prediction based on symptoms of diseases. This of AI is its ability to offer personalized and precise diagnoses
paper explores the application of machine learning in predicting and treatment plans by considering a patient’s unique med-
the disease related with symptoms. In constructing accurate
models capable of describing diseases, we make use of a range of ical history and other relevant data. This can lead to better
machine learning algorithms, such as support vector machines, treatment outcomes and reduced healthcare costs. Despite the
random forests, and decision trees. We prepare the data before potential benefits, there are challenges to using AI in medical
testing the models on various disease datasets and select the best disease identification. AI algorithms need to be thoroughly
features to improve the models. The results show that ML can tested to ensure they are accurate and reliable. There are also
accurately diagnose diseases much earlier. However, there are
still problems such as biases in algorithms, the quality of data, concerns about data privacy, security, and bias that need to be
as well as the performance of the model in most scenarios. This addressed to ensure AI is used safely and ethically. As research
research shows how ML can improve diagnostics in healthcare in this field progresses, it’s important to weigh the benefits and
and enhance disease prediction based on symptoms. risks carefully and use these new technologies responsibly.
Index Terms—Machine Learning,Naive Bayes One significant advantage of AI in disease identification is
Classifier,Random Forest,Support Vector Machines,Decision
Trees. its ability to quickly and efficiently process large amounts of
data, which is especially useful in areas with a shortage of
healthcare professionals or limited resources. AI can also help
I. I NTRODUCTION
doctors make more accurate diagnoses, especially for rare or
Medical disease identification is the process of finding out complex diseases. By analyzing data from different sources,
what disease a patient has based on their symptoms, medical AI can identify patterns and make predictions that might not
history, and other important information. This is a crucial be obvious to human experts. Besides improving diagnostic
task in healthcare because accurately diagnosing a disease is accuracy, AI can help identify new disease markers and risk
essential for providing the right treatment and improving a factors, which can enhance our understanding of how diseases
patient’s health. [1]. Traditionally, doctors and other healthcare develop and aid in the creation of new treatments. AI can
professionals use their knowledge and tools like medical also be used to monitor disease progression and treatment
imaging and lab tests to diagnose diseases. However, this effectiveness by analyzing changes in patient data over time,
can be time-consuming, expensive, and sometimes mistakes allowing for early intervention if problems arise. However,
occur. With the rise of artificial intelligence (AI), there is AI in medical disease identification also has limitations. AI
growing interest in using AI to help identify diseases. AI can algorithms need to be well-trained and validated to be sure
analyze large amounts of medical data to find patterns and they work well. There is also privacy and security concerns as
make predictions about diagnosis, treatment, and outcomes. well as having the potential for bias in decision-making. As
This can support traditional methods, making the process AI continues to develop in this area, then it will be important
faster and more accurate [2], [3]. AI-based medical disease to use it responsibly and ethically [6]–[9].
identification is still in its early stages, and there are challenges
to overcome, such as ensuring the accuracy and reliability of 1) Detect diseases early, increasing the chances of success-
AI algorithms. However, as this technology improves, it could ful treatment and preventing the spread of infections.
significantly change healthcare and lead to better outcomes 2) Ensure a correct diagnosis to provide appropriate treat-
for patients.The increasing use of electronic medical records, ment and avoid misdiagnosis, which can lead to ineffec-
medical imaging, and genetic data has created a need for tive treatments or serious health consequences.
faster and more accurate ways to identify diseases. AI can 3) Support public health efforts by identifying and tracking
address some of the limitations of traditional methods by diseases to prevent outbreaks.
analyzing data systematically and automatically. AI can learn 4) Contribute to research by providing accurate data that
from large amounts of data and use this knowledge to make can lead to new treatments and cures [10].
predictions, improving the speed and accuracy of diagnoses
[4], [5]. There are different types of AI algorithms that can be Machine learning (ML) is a type of technology where
used for disease identification, such as machine learning, deep computers learn from data to make decisions or predictions
learning, and natural language processing. These algorithms without needing detailed instructions for every task. It’s like
can be trained using various data sources, including medical teaching a computer by showing it examples [11].
A. How it Works 1) Healthcare: Machine learning helps doctors predict dis-
Data is the core of machine learning. Imagine you have tons eases, diagnose patients, and even suggest treatments. By
of data images of dogs and cats. You would like the computer analyzing medical records, it can identify patterns that may
to differentiate pictures of dogs from the pictures of cats. When indicate early signs of illnesses.
you have provided the computer with lots of examples of each, 2) Finance: Machines are used in the detection of illegal
it eventually learns what makes a cat different from a dog by transactions by banks, credit risk assessment, and even invest-
seeing patterns in the data [12], [13]. ment. It can analyze a thousand transactions in just a second,
making it identify strange patterns that may indicate fraud.
B. Types of Machine Learning 3) Retail: Online shops use machine learning to make
There are several types of machine learning, but the three recommendations for products you might like, based on what
main ones are: you have bought and browsed in the past. It is the technology
1) Supervised Learning:: This is how you are teaching the behind the ’customers who bought this also bought that’
computer what it should do for all these cases by providing suggestions.
labeled data. For example, you might give it a list of emails 4) Self Driving Cars: These cars apply machine learning
and classify a few of them as ”spam” and the others ”not to understand their surroundings. Based on data acquired
spam.” These instances teach the computer how to rank if an from cameras and sensors, cars can ’see’ the road, recognize
email that appears is spam or not [13]. obstacles, and make choices like a human [16], [17].
2) Unsupervised Learning:: In this case, the computer
operates independently. The input there is data with more
labels or expectations but no output of a specific response.
The machine then perceives patterns or trends in the data.
For example, without anyone instructing it as to the type of
information to search for, it may look through many consumer
purchases and discern patterns of different sorts of expenses
[14].
3) Reinforcement Learning:: This is like how animals learn
through rewards and punishment. The computer bases its
decision according to its judgments and receives feedback.
A reward is given to it for good decision-making, and for
bad, a penalty is issued. It acquires the ability to make better
decisions so as to maximize rewards over time [15].

Real World Example [17]

D. Challenges in Machine Learning
Machine learning is by no means weak, but it has its own
set of disadvantages. Quality in data is one of the major cons.
If the training data is biased or inadequate, the computer
will simply make wrong predictions or judgments. The issue
of interpretability is another concern; sometimes, in critical
domains like health or law, a machine learning model may
not be traceable as to how it reached a certain conclusion
[18].
E. The Future of Machine Learning
All these directions would be deeply integrated into life in
people’s daily activities if such technology is going that far.
We could think of things as a good deal smarter, much more
Fig. 1. Machine Learning accurate in their predictions, and maybe even solutions yet to
[16] be thought of. However, we have to truly responsibly create
such technologies—caring, just, and generally for society’s
good. In simple words, machine learning is just an amazing
C. Application of Machine Learning technique by which computers learn from data and exhibit
Machine learning is used in many areas today. Here are a decisions that improve the productivity and efficiency of
few examples: numerous jobs in a wide variety of industries [19], [20].
 II. M ACHINE L EARNING IN D ISEASE PREDICTION

There are various machine learning models that can be

used for disease identification. Here are some commonly used
models :

A. Logistic Regression
Logistic regression is perhaps the simplest, though most
widely used, classification algorithm; it predicts the likelihood
that a specific binary outcome will occur given one or more
predictor variables. Logistic regression is a type of statistical
model that enables us to make predictions about what the
possibilities are for an outcome with only two possible results,
like ”yes” or ”no,” ”success” or ”failure,” etc. It’s used when
you want to understand the relationship between one or more
factors called independent variables and the outcome that can
be one of two options, called a binary outcome. For example, Fig.3 Decision Tree [21]
logistic regression could predict whether or not a patient has
a disease based on a number of factors, including age, weight, C. Random Forests
and results from blood tests. It does this by calculating the This is a machine learning algorithm procedure wherein
probability of a given outcome and then using a formula to random forest makes better predictions by combining multiple
determine with input data what decision is best for that action choice trees. That is to say, consider asking a panel of experts
[21]. to have their say before making a decision rather than relying
on one person. Here is how it works: The random forest creates
many different decision trees, all of which are slightly different
because each uses different parts of the data. When it is time
to make a prediction, all the trees in the ”forest” vote, and the
most common result is given as the final answer. This works
well because aggregating the outputs of many trees gives a
more accurate and less likely wrong prediction than using
a decision tree. Random forests particularly work well with
complex data and are applicable to both number predictions
(regression) and category predictions (classification) [22].

Fig. 2. Logistic Regression [35]

B. Decision Trees
A decision tree is very easy to apply to data, looking vaguely
like a tree. Each part—or branch—of the tree represents a
decision or question about the data, and so each answer keeps
leading to another question or decision until you reach the
final result—a ”leaf” at the end of a branch. Imagine what it
must be like to be in that position—you’re trying to decide
whether to venture out. The first question probably will be,
”Does it shine?” If the answer is yes, then the next question
probably will be, ”Does it shine too hot?” If no, you probably Fig.4 Random Forests [22]
go out to play outside. If any answer is different, then you
go along a different path of questions until you come to D. Support Vector Machines
that final question. Decision trees help you make choices A support vector machine is a way in which the computer
by breaking down a complex problem into smaller, more separates data into two categories. So, you are given dots on
manageable questions [21]. a piece of paper, but some are red and the rest are blue.
A support vector machine is going to try and find the best
possible straight line (or boundary) that could separate the red
dots and the blue dots so each of the dots is on opposite sides
of it. Therefore, it becomes the object that the boundary must
be drawn far enough away from the dots of both colors. It
does make sure that even if you will draw up new dots, the
strength and reliability of the boundary are ensured for such
drawings. The SVM works well for more complex situations
than just dots on a piece of paper, but perhaps the data is larger
or in more dimensions or more difficult to separate. For this
kind of situation, an SVM applies special mathematical tricks
for delimiting boundaries in such a way that still works when
deciding between different groups [23], [24].
Fig.6 Artificial Neural Network [25]

F. Convolutional Neural Networks (CNNs)

A CNN is an artificial neural network that is particularly
remarkable at identifying features in images and recognizing
the pattern. It operates similarly to an eye of a human by
detecting appropriate features, such as shapes and edges, for
a correct interpretation of what an image shows. The network
consists of layers that function as filters that scan through
the picture in search of these features. In the first case, it
somehow starts from trivial patterns—lines—and combines
them in order to recognize more complex details like eyes
or wheels. CNNs are widely applied in such tasks as image
object detection, face recognition, and medical image analysis
[27].

Fig.5 Support Vector Machine [23]

E. Artificial Neural Networks (ANNs)

An artificial neural network is a computer system being Fig.7 Convolutional Neural Network [27]
designed to work like the human brain. It’s made of layers of
G. Recurrent Neural Networks (RNNs)
connected units called ”neurons” that process information. The
first layer receives data, the middle layers learn patterns in the A recurrent neural network is a type of computer program
data, and the last layer decides or predicts what the program designed to understand data that comes in a sequence, such
thinks should happen. Every connection between neurons has as sentences or time series. Unlike ordinary networks, RNNs
a weight that indicates how important it is to the network. have loops, which means they can remember what happened
During training, the network alters these weights so that it will before. It is great in handling tasks where past information is
become good at predicting things more accurately. ANNs are crucial, such as predicting the next word of the sentence or
widely applied to many applications such as face recognition, when analyzing video frames. It is essentially like having a
understanding speech, and recommending items on websites memory in RNNs, which helps learn patterns in time. This is
[25], [26]. what enables the network to make more informed decisions
based on current data as well as previous data. RNNs are often
used in language translation, speech recognition, and music
generation [28].
 those like blue, then you would guess that the new person
would like blue too. KNN works like that by using distance
to find the closest examples in its data to make a prediction or
classification. Very simple, though best with smaller datasets
[29], [30].

Fig.8 Recurrent Neural Network [28]

H. Naive Bayes
Naive Bayes is a simple algorithm that allows computers to
make predictions or classify things based on their probability. Fig.10 K-Nearest Neighbour [30]
It’s based on Bayes’ Theorem, and it assumes that all of your J. Gradient Boosting Machines (GBMs)
features are independent. In reality, of course, they aren’t, but
that’s why it’s called ”naive.” It works surprisingly well in A computer uses Gradient Boosting Machines (GBM) to
many situations, like, as everyone who has an e-mail account make better predictions by combining the outputs from many
knows, spam email detection or classifying news articles. It smaller models that are not so strong into one strong model.
works by assuming past data to calculate the probability of This happens by first making a simple prediction, then looking
different outcomes and making predictions from them. It is a at where it went wrong and building a new model to correct
fast algorithm, easy to use, and can handle large datasets well. those errors. With each step, it improves its whole prediction.
Its popularity in basic classification tasks is attributed to this With each new model, it focuses on the mistakes of the
[28]. preceding one, thereby becoming perfect gradually. The result
is a very powerful model that can deal with complex data
and make very accurate predictions, especially for tasks like
classification and regression [31].

Fig.9 Naive Bayes [28] Fig.11 Gradient Boosting Machines [31]

I. K-Nearest Neighbors (KNN) K. Long Short-Term Memory (LSTM)
KNN -K Nearest Neighbors: This is a simple way for Long Short-Term Memory is a subtype of artificial neural
computers to make predictions or classify things on the basis network that enables computers to remember certain informa-
of data they already have. You have a group of people tion over a long period of time, especially in tasks such as
whose favorite color is known. In that case, you meet a new understanding sentences or predicting future events. Unlike
person. To guess their favorite color, you look at their ”nearest regular neural networks that can quickly forget important
neighbors”—the” people most similar to them. If most of details, LSTMs are designed to remember the key information
over longer periods and then forget the unimportant details. own pros and cons, and their performance can vary depending
This therefore makes them better at sequences such as speech, on the specific disease. But their review doesn’t go into much
text, or time series data. LSTMs have become common in detail about how these algorithms are applied in real life
tasks such as language translation, speech recognition, and situations . [2] Arumugam studied how machine learning can
predicting stock prices [32]. predict heart disease in people with diabetes. They found that
Decision Tree models worked better than Naive Bayes and
Support Vector Machines (SVM) for this task. They noted
that the model still needs some adjustments to improve its
accuracy. Their findings might not apply to everyone because
they only looked at diabetic people. [3]Aldahiri looked at how
combining IoT (Internet of Things) with machine learning
can improve healthcare. They focused on how real-time data
analysis helps make better decisions and predict diseases more
accurately. However, they also pointed out that the results can
vary depending on the data used, and it can be difficult to keep
the IoT data consistently high-quality . [4] It explores how
Fig.12 Long Short-Term Memory [32] machine learning can predict diseases based on symptoms,
L. Extreme Gradient Boosting (XGBoost) highlighting the importance of early detection. It showed that
predicting diseases using symptoms can be effective, but their
XGBoost, or Extreme Gradient Boosting, is one of the
study mainly focused on specific models like Naive Bayes
strongest prediction-making methods when operating off data.
and Decision Tree, which might not work for all diseases.
How it works is by simply trying to grow a sequence of
They also mentioned possible problems with the accuracy and
decision trees, where each subsequent tree attempts to correct
reliability of these predictions . [5] Machine learning can find
the errors of its predecessor. Imagine how to improve a recipe
connections between diseases and treatments in medical texts.
by adding small tweaks each time. XGBoost generates the
Their method successfully identifies important relationships,
trees one by one in a manner more aggressive in areas where
helping to share medical knowledge more effectively. How-
previous trees made mistakes. Thus, with each step, it works
ever, they also pointed out that the quality and completeness
better and more precisely. It is fast, suitable for large datasets,
of the medical texts, as well as the challenge of capturing more
and often participates in competitions of machine learning for
subtle connections, can be limiting factors. [6]Deo reviewed
high performance [33], [34].
how machine learning could be used in medicine, discussing
its potential benefits and challenges. The review showed that
while machine learning has a lot of promise, its impact on
actual medical care is still limited. This is mainly because of
problems with data quality, difficulties in integrating it into
healthcare systems, and the need for more clinical testing and
validation . [7] Kumar looked at how machine learning and
deep learning can help diagnose brain diseases like glioblas-
toma. They introduced models like GLCM and linking metric
models, which improve the accuracy of diagnoses. However,
these methods require a lot of computing power and may not
work as well for other diseases or different types of medical
imaging . [8] It developed a system that uses multiple machine
learning algorithms to predict diseases based on symptoms.
The weighted KNN algorithm performed well, achieving a
high accuracy of 93.5, showing that the system is effective for
Fig.13 Extreme Gradient Boosting [34] early disease diagnosis. However, they noted that the system’s
performance might change depending on the quality of the
III. L ITERATURE R EVIEW data and the variety of diseases included. [9] Reddy created
Recent research highlights the growing application of ma- a system that predicts diseases based on symptoms, making it
chine learning (ML) in disease prediction and diagnosis, with easier for people to get an early diagnosis. While the system is
a focus on various algorithms and their effectiveness across helpful for initial diagnoses, its accuracy and reliability depend
different healthcare contexts. [1]Ibrahim reviewed different on the quality of the data and the specific machine learning
machine learning algorithms used for diagnosing diseases. model used. It also might not cover all possible diseases .
They looked at Naive Bayes, logistic regression, SVM, K- [10] Dahiwade proposed a disease pre diction model using
nearest neighbor (KNN), K-means clustering, Decision Tree, KNN and Convolutional Neural Networks (CNN). They found
and Random Forest. They found that each algorithm has its that CNN performed better than KNN, with an accuracy of
84.5. While CNN is efficient in terms of time and memory,
it needs a lot of computational power to work effectively.
[11]It discussed how machine learning can improve disease
diagnosis and prognosis by using large datasets. They pointed
out that while machine learning can be very helpful for early
disease detection, the early results might not yet be fully tested
or confirmed in real world clinical settings. [12] Pahul Preet
Singh Kohli looked at different machine learning classification
algorithms for predicting diseases using datasets from the UCI
repository. The paper supports using machine learning for
early disease detection but might not cover all the real-world
factors that can affect how accurate the predictions are. [13]
This Paper reviewed how machine learning works with IoT Fig.14 ML Algorithm [35]
(Internet of Things) in healthcare. They found that machine
learning improves how IoT systems function, but they also IV. D ISCUSSION
pointed out concerns about security and privacy. The review
might not include all the latest advancements in this area. The reviewed papers give an extensive view of how machine
[14] Dr. Hemsagayathri looked at computer-aided diagnostic learning is applied in the prediction and diagnosis of diseases,
systems that use machine learning to detect diseases based showing both the progress accomplished and the ones remain-
on symptoms. They found that these systems greatly improve ing to be done. Here’s a summarized version of the key points
diagnostic accuracy. However, they also noted that these mod- in simple terms:
els have limitations when it comes to dealing with symptoms
that overlap or are similar to one another. [15] Sneha Gram A. How Well Do Machine Learning Algorithms Work?
Purohit studied different machine learning algorithms, such as
Decision Tree, Random Forest, and NaiveBayes, for predicting 1) Strengths:: Different Algorithms for Different Tasks:
diseases. The paper highlights how well these algorithms Different ML algorithms, like decision trees, naı̈ve Bayes,
work but also notes that there are limitations related to the and KNN, are found to be more effective in the prediction
specific datasets used and the need for data preprocessing. of different diseases. For example, Arumugam et al. (2021)
[16] Md. Ahsan reviewed machine learning techniques for mentioned that their decision tree algorithm was the best as
diagnosing diseases, covering recent trends and challenges. compared to Naive Bayes and SVM for the prediction of heart
They highlighted the increasing use of machine learning but disease in diabetes patients. Similarly, Keniya et al. (2021)
may not have fully covered the practical challenges involved reported that their KNN-based algorithm performed very well
in putting these techniques into use. [17] K. Gaurav and for the prediction of diseases [36].
colleagues explored machine learning techniques like Random Advanced Models for Complex Diseases: Models like
Forest, SVM, and LSTM for predicting diseases using real- Convolutional Neural Networks (CNNs) and deep learning, as
life data. They found that these techniques improve accuracy discussed by Kumar et al. (2021) and Dahiwade et al. (2021),
but also noted some limitations related to predicting based on are quite appropriate for complex conditions like brain tumors.
symptoms and the quality of the data. [18]This Paper explains These advanced models yield better predictions in comparison
about different machine learning algorithms, like Naive Bayes, to the traditional methods.
Decision Tree, and Linear Regression, to predict diseases from 2) Limitations:: Dependence on the Disease and
symptoms. The paper highlights the potential for better early Data:Obviously, the performance of ML algorithms strongly
detection, but there might be challenges when applying these depends on the specific disease and even on the quality of the
methods in real-life situations. [19] H. Asadi reviewed how data. This underlines the need for the choice and fine-tuning
machine learning can advance medical practice by handling of the right model for each particular case. [37].
complex and large-scale data. The review gives an overview Need for Resources and Quality Data: Advanced models
of the positive impact of machine learning but might not require significant computing power and high-quality data to
cover the practical challenges of using these methods in real- function effectively. Issues related to data quality and the
life settings. [20] M.A. Majid explored how AI and machine necessity for thorough testing can limit their practical use.
learning are used for predicting diseases, discussing different
methods, applications, and future possibilities. The chapter B. Combining ML with IoT
highlights the potential of AI and ML but also notes challenges
related to data quality, privacy, and biases in algorithms. This 1) Strengths:: Better Decision-Making:The integration of
summary captures the contributions and limitations of each ML with the Internet of Things (IoT) could allow for improved
paper, highlighting the ongoing advancements and challenges decision-making by using real-time data. It may increase the
in applying ML to disease prediction and diagnosis. accuracy in the prediction of diseases as well as support timely
interventions [38].
 2) Limitations:: Data Quality and Privacy:Maintaining need for large datasets, and ensuring predictions are reliable
high data quality while ensuring privacy and security presents and unbiased. Ongoing research and advancements in these
ongoing challenges. These issues must be managed carefully areas are essential to fully leverage the potential of machine
to ensure the system operates effectively and protects patient learning in healthcare [44]–[46].
information [38].
R EFERENCES
C. Predicting Diseases from Symptoms [1] The Role of Machine Learning Algorithms for Diagnosing Diseases by
1) Strengths:: Early Detection:ML models that predict Ibrahim and Abdulazeez (2021).
[2] Multiple Disease Prediction Using Machine Learning Algorithms by K.
diseases based on symptoms can be highly beneficial for early Arumugam ,Mohd Naved,Priyanka P.Shinde(2021).h
diagnosis. This aids in detecting diseases early and initiating [3] Trends in Using IoT with Machine Learning in Health Prediction
treatment sooner [38]. Systems by Aldahiri ,Bashir and Hussain (2021)
[4] Disease Prediction Using Machine Learning Techniques Mallela ,Bha-
2) Limitations:: Accuracy Issues: These models can some- vani and Lakshmi (2021)
times struggle with accuracy, particularly when symptoms are [5] A Machine Learning Approach for Identifying Disease-Treatment Rela-
similar or overlapping. Careful data preparation and model tions in Short Texts Frunza ,Thomas,MemberIEEE (2021)
[6] Machine Learning in Medicine Deo (2015)
selection are necessary to enhance their reliability [39]. [7] Machine Learning and Deep Learning Approaches for Brain Disease
Diagnosis: Principles and Recent Advances Kumar et al. (2021)
D. Handling Large Datasets and Complex Data [8] Disease Prediction from Various Symptoms Using Machine Learning
Keniya et al. (2021)
1) Strengths:: Improved Diagnostic Capabilities: Ni- [9] Disease Predictor Based on Symptoms Using Machine Learning Reddy
harika G. Maity et al. (2017) and H. Asadi et al. (2019) et al. (2021)
have highlighted that the benefit of using big data and proper [10] Designing Disease Prediction Model Using Machine Learning Approach
Dahiwade et al. (2021)
handling of complex data helps bring about better diagnostic [11] Machine Learning for Improved Diagnosis and Prognosis in Healthcare
ability. Since ML can analyze a large amount of data, it could Niharika,Sreerupa Das (2017)
reveal patterns or perspectives missing out from traditional [12] Application of Machine Learning in Disease Prediction Pahulpreet Singh
et al.(2018).
data analyses. [40]. [13] A Review on the Role of Machine Learning in Enabling IoT
2) Limitations:: Practical Challenges: There are various Based Healthcare Applications HEMANTHA KRISHNA BHARAD-
challenges associated with data quality, integration into ex- WAJ et.al(2021).
[14] Symptoms Based Disease Prediction Using Machine Learning Tech-
isting healthcare systems, and real-world validation, which, niques by P.Hesagayathri .et.al(2021).
according to Deo (2015) and Ahsan et al. (2022), need to be [15] Disease Prediction using Machine Learning Algorithms by Sneha
addressed in order to translate the accomplishments of ML et.al(2020).
[16] Machine-Learning-Based Disease Diagnosis: A Comprehensive Review
into tangible clinical benefits [41]. by Md Manjurul Ahsan et.al(2022).
[17] Human Disease Prediction using Machine Learning Techniques and
E. Overall Trends and Future Directions Real-life Parameters by K. Gaurav, A. Kumar, P. Singh, A. Kumari,
1) Strengths:: Growing Potential:Research indicates that M. Kasar, T. Suryawanshi.(2023).
[18] Disease Prediction using Machine Learning by Kedar Pingale et.al(2019)
ML and AI possess significant potential in healthcare, with [19] eDoctor: machine learning and the future of medicine by G. S. Handel-
advancements in algorithms and data processing paving the man et.al(2018).
way for improved disease prediction and diagnosis [42]. [20] Book: Contemporary Applications of Data Fusion for Advanced Health-
care Informatics by G.S. Karthick PSG College of Arts and Science,
2) Limitations:: Persistent Challenges:Nonetheless, the India Sathishkumar Karupusamy Gobi Arts and Science College, India
problems of data quality, computational resources, and system (2023).
integration are still faced. Testing the ML model in real-world [21] Google Scholar and Wikipedia
[22] Shaik Razia, P.Swathi Prathyusha, N.Vamsi Krishna, N.Sathya Sumana,
situations and considering privacy issues are to be ensured and ”A Review on Disease Diagnosis Using Machine Learning Techniques,”
realized for its massive adoption [43]. International Journal of Pure and Applied Mathematics, Volume 117, No.
16, 2017.
V. CONCLUSION [23] Diyar Qader Zeebare, Habibollah Haron, Adnan Mohsin Abdulazeez
and Dilovan Asaad Zebari, ”Machine learning and Region Growing for
In our review of disease prediction using machine learn- Breast Cancer Segmentation”, International Conference on Advanced
ing, we found that these techniques are highly effective for Science and Engineering, IEEE, 2019.
[24] Hossam Meshref, ”Cardiovascular Disease Diagnosis: A Machine Learn-
predicting various diseases. They can quickly analyze large ing Interpretation Approach”, International Journal of Advanced Com-
datasets and identify patterns that might not be apparent to puter Science and Applications, Vol. 10, No. 12, 2019.
humans. Algorithms such as decision trees, neural networks, [25] Diyar Qader Zeebaree, Habibollah Haron, Adnan Mohsin Abdulazeez,
and Dilovan Asaad Zebari, ”Trainable Model Based on New Uniform
and support vector machines have demonstrated success in LBP Feature to Identify the Risk of the Breast Cancer”. In 2019 Interna-
predicting diseases like diabetes, heart disease, and cancer. tional Conference on Advanced Science and Engineering (I.C.O.A.S.E.)
However, prediction accuracy heavily depends on data quality, (pp. 106-111). IEEE, April 2019.
[26] Iswanto Iswanto, E. Laxmi Lydia, K. Shankar, Phong Thanh Nguyen,
algorithm choice, and model training. Machine learning holds Wahidah Hashim, Andino, ”Identifying Diseases and Diagnosis using
significant promise for enhancing disease prediction, making Machine Learning”, International Journal of Engineering and Advanced
it faster and more accurate. This technology can facilitate Technology, Vol. 8, August 2019.
[27] Reem A. Alassaf, and et al, ”Preemptive Diagnosis of Chronic Kidney
early detection, which is critical for improving treatment Disease Using Machine Learning Techniques”, International Conference
outcomes. Yet, challenges remain, including data quality, the on innovations in Information Technology (I.T.), IEEE, 2018.
[28] Sneha Grampurohit and Chetan Sagarnal, ”Disease Prediction using
Machine Learning Algorithms” International Conference for Emerging
Technology (I.N.C.E.T.), Belgaum, India. Jun 5-7, 2020.
[29] Divya Jain and Vijendra Singh, “Feature selection and classification
systems for chronic disease prediction: A review”, Elsevir, 2018.
[30] Nikita Banerjee Subhalaxmi Das, Prediction Lung Cancer– In Machine
Learning Perspective, IEEE, 2020.
[31] Uday Pratap Singh, Siddharth Singh Chouhan, Sukirty Jain, And Sanjeev
Jain”, IEEE, 2019
[32] A.Deva Kumar, Josephine Prem Kumar, V.S Prakash and Divya K.S,
”Supervised Learning Algorithms: A Comparison”, Kristu Jayanti Jour-
nal of Computational Sciences, Vol. 1, Issue 1, pp. 01-12, 2020.
[33] Autsuo Higa, “Diagnosis of Breast Cancer using Decision Tree and Ar-
tificial Neural Network Algorithms”, International Journal of Computer
Applications Technology and Research, Vol. 7, Issue 01, pp.23-27, 2018.
[34] D. Selvathi and K. Suganya, ”Support Vector Machine Based Method for
Automatic Detection of Diabetic Eye Disease using Thermal Images”,
2019
[35] Rahma Atallah and Amjed Al-Mousa, ”Heart Disease Detection Using
Machine Learning Majority Voting Ensemble Method”, IEEE, 2019
[36] Nikita Banerjee Subhalaxmi Das, Prediction Lung Cancer– In Machine
Learning Perspective, IEEE, 2020.
[37] N. Komal Kumar, G.Sarika Sindhu, D.Krishna Prashanthi, A.Shaeen
Sulthana, ”Analysis and Prediction of Cardio Vascular Disease using
Machine Learning Classifiers”, 6th International Conference on Ad-
vanced Computing and Communication Systems (I.C.A.C.C.S.), 2020.
[38] Mahzabeen Emu, Farjana Bintay Kamal, Salimur Choudhury and Thiago
E. Alves de Oliveira, ”Assisting the Non-invasive Diagnosis of Liver
Fibrosis Stages using Machine Learning Methods”, IEEE, 2020.
[39] Dayanand Jamkhandikar and Neethi Priya, ”Thyroid Disease Prediction
Using Feature Selection And Machine Learning Classifiers”, The Inter-
national Journal of analytical and experimental modal analysis, ISSN
NO: 0886-9367, 2020.
[40] Vidya M and Maya V Karki, ”Skin Cancer Detection using Machine
Learning Techniques”, IEEE, 2020.
[41] Senthil Kumar Brindha and et al, ”Data Mining for Early Gastric Cancer
Etiological Factors from Diet-Lifestyle Characteristics”, Proceedings
of the International Conference on Intelligent Computing and Control
Systems, IEEE, 2020
[42] Halgurd S. Maghdid, Aras T. Asaad, Kayhan Zrar Ghafoor, Ali Safaa
Sadiq and Muhammad Khurram Khan, “Diagnosing COVID-19 Pneu-
monia from X-Ray and CT Images using Deep Learning and Transfer
Learning Algorithms”, arXiv preprint arXiv:2004.00038, 2020
[43] Md. Zabirul Islam, Md. Milon Islam and Amanullah Asraf, “A com-
bined deep CNN-LSTM network for the detection of novel coronavirus
(COVID-19) using X-ray images”, Elsevier, 2020
[44] Xiaolu Tian and et al, ”Using Machine Learning Algorithms to Predict
Hepatitis B Surface Antigen Seroclearance”, Hindawi Computational
and Mathematical Methods in Medicine Volume, 2019
[45] Adel Sabry Eesa, Zeynep O.R.M.A.N., Adnan Mohsin Abdulazeez ,
”A new feature selection model based on ID3 and bees algorithm for
intrusion detection system”, Turkish Journal of Electrical Engineering
Computer Sciences, 2015.
[46] Yash Jayesh Chauhan, ”Cardiovascular Disease Prediction using Clas-
sification Algorithms of Machine Learning”, International Journal of
Science and Research (I.J.S.R.), ISSN: 2319-7064, 2018.