Journal of Artificial Intelligence and Metaheuristics (JAIM) Vol. 07, No. 02, PP.

18-31, 2024

Advancements and Future Directions in Machine Learning

for Medical Diagnostics: A Comprehensive Review

Basant Sameh1 , Nima Khodadadi2, Ehsan khodadadi3, Marwa M. Eid4, S. K. Towfek5

1
Department of Communications and Electronics, Delta Higher Institute of Engineering and Technology
(DHIET), Mansoura 35111, Egypt
2
Department of Civil and Architectural Engineering, University of Miami, Coral Gables, FL, USA
3
Department of Chemistry and Biochemistry, University of Arkansas, Fayetteville, AR 72701, USA
4
Faculty of Artificial Intelligence, Delta University for Science and Technology, Mansoura 35111, Egypt
5
Computer Science and Intelligent Systems Research Center, Blacksburg 24060, Virginia, USA
Emails: [email protected]; [email protected]; [email protected]; [email protected];
[email protected]

Abstract
Machine learning (ML) based techniques have enjoyed significant popularity in addressing the hostility of
numerous problems in a range of applications, such as finance, marketing, production, environment, health care,
and security. One of the most important distinctions between machine learning and human ways of thinking is
their ability to observe patterns, make interpretations, reveal some hidden relationships, and analyze huge amounts
of data. Machine learning (ML) technology can lead to improved specificity, sensitivity, predictability, and
steadiness of such systems. Through this review, though, we will have an in-depth discourse on the application of
machine learning in the field of medicine and how the latest technologies are mostly deployed in diagnostics.
Medical applications that are widely used, including but not limited to machine learning solutions for medical
chemistry, wearable sensors, cancer, the brain, and medical imaging, will be discussed in detail, with a focus on
model adjustments to address the problems faced by the applications. In the course of the work, academics,
practitioners, and decision-makers will have plenty of opportunities to utilize the findings, references, and insights
of this study to improve their work and steer future research.
Keywords: Machine learning; Artificial Intelligence; Machine Learning Applications; Medical Field; ML in
Healthcare.
1. Introduction

Machine learning (ML), also known as artificial intelligence (AI), though frequently falsely confused, stands for
a separate discipline widely applied in various disciplines, including healthcare. Smart diagnostics based on
machine learning methods are the real, invaluable solutions that sensing technologies can leverage in the healthcare
domain, such as in medical imaging or cancer diagnosis [1]. By using machine learning algorithms, there is an
opportunity for the medics to tackle the critical indicators of medical data extraction, diagnosis, and progression
of the disease processes. This, in turn, serves to facilitate informed decisions and optimize patient care pathways.
In addition to that, the concept of machine learning is of huge importance because it provides healthcare systems
with the skills to speed up the data collection processes, which therefore makes possible the timely detection and
dissemination of informative alerts and the proactive implementation of the required interventions [2].

Also, machine learning applications in specialized hospital care involve the careful examination of medical data
to formulate patterns and derive useful information that they can use. This invites the respective encodings of the
patient information so that the algorithms can be matched with the learning goals. This should be done to enhance
diagnostic precision and relevancy. The machine learning umbrella encapsulates a multitude of scenarios whereby
18
DOI: https://doi.org/10.54216/JAIM.070202
Received: May 17, 2023 Revised: September 19, 2023 Accepted: February 17, 2024
 Journal of Artificial Intelligence and Metaheuristics (JAIM) Vol. 07, No. 02, PP. 18-31, 2024

each situation requires supreme attention to detail in order to guarantee the use of classifiable data and to match
certain diagnostics. By providing technicians with the capability to collect the data along with issues that are
medically related to each other, machine learning makes sure healthcare professionals get quick, precise, and error-
free diagnoses even with multiple and newly diagnosed cases. Lastly, the widespread availability and simpleness
of tools for diagnostics empowered by machine learning are thus not just for healthcare experts but also for students
and those who are not in the medical profession, which in turn opens the avenues for many fresh people in this
field and helps with better medical outcomes.

Besides, the explanatory, detailed understanding of the fundamental techniques, methodology, and concepts of
machine learning and their real-world applications in the medical field uncover their transformative nature in
healthcare delivery. Integrating an integrated aspect of the literature review, this study aims to illuminate the core
principles, advanced utilization and eventual applications of machine learning in medical diagnosis, thereby
advancing disease detection, personalized treatment strategies and improved patient outcomes. With the prudent
integration of machine learning algorithms, the health sector enters the next phase of a data-driven approach, which
leads to the rise of precision medicine and predictive healthcare paradigms.
2. Literature Review
The development of machine learning applications (ML) in medicine has witnessed a significant breakthrough,
primarily because of the wide breadth of data availability, processing power, and algorithmic cleverness that have
emerged. ML techniques show promising potential not only for speeding up but also for remodeling healthcare
delivery. Emerging applications range from diagnosis and prognosis to treatment planning and personalized
medicine. This literature review aims to cover all the medical areas with a wide application of ML techniques. We
also analyze issues and changes related to the implementation of ML into clinical practice. Through analysis of
the recent research, methodologies and results, this review attempts to clarify the formalizing area of ML that uses
those technologies in the diagnosis of diseases and advanced treatment and generally contributes to the
transformation of healthcare.
2.1. Summary Table
As shown in Table 1, The literature review concludes that the level of machine learning influence in the medical
domain is expected to grow and drive a paradigm shift that will cause medical care to be perceived in a new way.
The findings of the existing research illustrate how machine learning enhances diagnostics, prognosis, and
treatment management in healthcare. Indeed, therein lies the promise, but interpretability, ethics, and the accuracy
of validation are still being worked on. With ML applications still growing and advancing, a corresponding bridge
between innovation and practical implementation is, to date, a topical and most crucial aspect. The review shows
how machine learning can change medical practices, saying that the researchers, the practitioners, and the
policymakers need to partner to work through the dynamics of the evolving landscape in favor of the patient's care
and the results.
Table 1: Summary of related works.

Ref. Focus Methodology Key Findings

[3] Machine Learning Overview and survey of ML - ML's popularity in handling issues in health, finance,
Applications in Various applications marketing, security, medicine, and ecology. - ML's unique
Fields abilities in pattern discernment, data interpretation, and
large dataset analysis. - ML's potential for improving
accuracy, reliability, effectiveness, and anticipative ability
in disease diagnosis. - Primary influence of mature
technology on medical diagnosis. - Exploration of five
broad medical applications: radiology, neurology,
pharmacology, cancer, and general medicine.
[4] Risks of AI Applications in Systematic review (1989- - Examination of architectural design in ML/radiomics
Medical Decision Support 2021) of ML/radiomics and and DL methods. - Core processes explored: option
DL methods selection, training, validation, and testing. - Role of
convolutional neural networks in direct image processing.
- Focus on key tech procedures for data curation: image
labeling, annotation, data harmonization, and federated
19
DOI: https://doi.org/10.54216/JAIM.070202
Received: May 17, 2023 Revised: September 19, 2023 Accepted: February 17, 2024
 Journal of Artificial Intelligence and Metaheuristics (JAIM) Vol. 07, No. 02, PP. 18-31, 2024

learning. - Consideration of challenges in choosing

between ML and DL for medical imaging.

[5] Evolution of Machine Review of recent systematic - ML's role in uncovering complex medical data. -
Learning in Medicine analysis in medical data Emphasis on the needful use of growing medical data in
management healthcare fields. - Review of versatile ML methods in
multiple medical arenas. - Evaluation of the shift from ML
to DL in recent medical data management.
[6] AI Integration in Medical Comprehensive examination - Popularity of AI in medical professions like radiology,
Imaging of AI principles in medical pathology, and cancer screening. - Modernization of
imaging equipment and focus on research and development. -
Increasing adoption of AI in medical imaging for
diagnosis, segmentation, and classification. - Emphasis on
the need for knowledgeable practitioners for safe and
efficient AI utilization.
[7] Automating Machine Literature review on AutoML - Descriptive aspects of AutoML and its role in healthcare.
Learning (AutoML) in - Identification of future opportunities and drawbacks. -
Healthcare Current applications of AutoML in the industry. -
Highlight of AutoML's advantages in achieving
performance levels comparable to or higher than human-
produced results.
[8] ML and DL Techniques in Application of ML and DL - Importance of computer-aided detection in accurate
Breast Cancer techniques in breast cancer early detection of breast cancer. - Classification of breast
detection cancer using various imaging techniques. - Review of
different ML and DL models for breast cancer
identification. - Aim to improve diagnostic processes for
breast cancer.
[9] ML and AI in Pandemic Utilization of ML and AI in - Application in screening, predicting, forecasting, contact
Response pandemic response processes tracing, and drug design. - Evaluation of contributions
against SARS-CoV-2. - Recommendations for
researchers, model developers, and policymakers. -
Potential of AI and ML innovations to halt further
pandemics and reduce human intervention in healthcare
operations.
[10] Data-Driven Techniques in Examination of ML - Impact of data-driven techniques in life sciences. -
Scientific Domains techniques in life sciences Relevance of ML and computational intelligence in
situations with a wealth of data. - Concerns about
interpretability and explainability of nonlinear models. -
Call for integration of methodologies for readability in
healthcare decision-making.
[11] Application of ML and DL Review of 40 articles on ML - Exploration of ML and DL techniques used in medical
in Medical Imaging and DL applications in imaging. - Assessment methods and operations discussed.
medical imaging - Patterns for datasets considered. - Comparative analysis
of machine learning classifiers and deep learning models
using MRI dataset. - Goal to provide healthcare
community with tools for wiser diagnostic decisions.
[12] ML Applications in Comprehensive review of ML - Exploration of the influence of AI and ML in orthopedic
Orthopedics applications in orthopedics medical care. - Consideration of big data and its role in
improving musculoskeletal therapy. - Examination of the
combination of modern data analytics and orthopedic

20
DOI: https://doi.org/10.54216/JAIM.070202
Received: May 17, 2023 Revised: September 19, 2023 Accepted: February 17, 2024
Journal of Artificial Intelligence and Metaheuristics (JAIM) Vol. 07, No. 02, PP. 18-31, 2024

work. - Focus on the major impact of AI and ML

technologies in the field of orthopedics.

2.2 Review
The development of machine learning applications (ML) in medicine has witnessed a significant breakthrough,
primarily because of the wide breadth of data availability, processing power, and algorithmic cleverness that have
emerged. ML techniques show promising potential not only for speeding up but also for remodeling healthcare
delivery. Emerging applications range from diagnosis and prognosis to treatment planning and personalized
medicine. This literature review aims to cover all the medical areas with a wide application of ML techniques. We
also analyze issues and changes related to the implementation of ML into clinical practice. Through analysis of
the recent research, methodologies and results, this review attempts to clarify the formalizing area of ML that uses
those technologies in the diagnosis of diseases and advanced treatment and generally contributes to the
transformation of healthcare.
The machine learning (ML) technique has been one of the most popular methods for handling various difficult
issues that are widely found in specific fields such as health, finance, marketing, security, medicine, and ecology.
Machine learning stands out from other machine learning techniques because it can discern patterns, render
interpretation, indiscriminately analyze large datasets, and uncover constructive correlations between data points.
Machine learning (ML) is a tool with a high potential for improving the accuracy, reliability, effectiveness, and
anticipative ability in disease diagnosis in the medical domain especially. A particular report deals with medical
machine learning (ML) applications, and this overview covers the primary influence of mature technology on
medical diagnosis. [3] sheds light on five broad medical applications, namely, radiology, neurology,
pharmacology, cancer, and general medicine. This research brings great ideas to the market that benefit individuals,
companies, and societies in the future. Researchers, practitioners, and decision-makers must get this information
from the survey. In [4], AI models are frequently known as one of the most effective instruments in the face of
quickly developing molecular medicine and medical services. This investigation is about a specific topic:
identifying the risks of artificial intelligence applications, which are the decision support systems that doctors and
other qualified medical personnel often use to make important decisions in practical and real scenarios. This paper
does a systematic review to examine the works written between 1989 and 2021, focusing on the architectural
design of machine learning (ML)/radiomics and deep learning (DL) methods. These processes in machine learning
and radiomics, such as option selection, training, validation, and testing, have been at the core of this research.
Besides this, another story about how deep learning models work is that convolutional neural networks play a
responsible role in the process of direct image processing. The focus of this investigation is to delve into the
foundation of the key tech procedures that are needed for data curation, including the processes of image labeling,
annotation, data harmonization, and federated learning. Whether to choose machine learning or deep learning for
a medical imaging application is seen in these good and bad sides. This element comprises several factors,
including selecting the sample size, exploiting data augmentation for limited and unbalanced datasets, and the
interpretability of artificial intelligence models. The suggestion perceives the presence of these problems as an
impediment to introducing AI solutions to medical imaging and those of primary concern to clinical practice.
Machine learning, being one of the most insurmountable and exceedingly powerful tools among others, plays a
great role in the uncovering of complex medical data. The growing volume of medical data, coupled with the
previous explanation of its needful use for the world's medical and healthcare fields, is an urge-like sensation for
the data to be well-utilized. [4] first conducts an exhaustive review of the systematic analysis that was published
recently, mainly focusing on the utilization of versatile machine learning methods in multiple medical arenas. This
article initially discusses the progress achieved and then evaluates how machine learning and deep learning
methods and principles have recently been applied to the field of medical data management. According to this
study, the implementation of AI-based methods has brought a comparatively large change in medicine, where the
DL-based approach is gradually taking the place of the ML-based approach one by one in recent times. The term'
Artificial Intelligence' (AI) has been quite popular over the last few years as technology is changing day by day,
and we have some evidence to prove it. Some medical professions, e.g., radiology, pathology, and cancer
screening, have shown the potential the phenomenon can offer in the medical field with undeniable excitement.
They have modernized the equipment, which is of importance to research and development. As such, they have
focused on integrating artificial intelligence capabilities into medical clinics. As the field of medicine is
progressively adopting more and more artificial intelligence (AI) in medical imaging, including diagnosis,
segmentation, and classification, a knowledgeable practitioner should be chosen to guarantee that AI utilization is
done safely and efficiently. The goal of [6] is to comprehensively explain the principles of technological AI that
are considered to be the most advanced ones and how those principles can be applied in the medical imaging field
21
DOI: https://doi.org/10.54216/JAIM.070202
Received: May 17, 2023 Revised: September 19, 2023 Accepted: February 17, 2024
Journal of Artificial Intelligence and Metaheuristics (JAIM) Vol. 07, No. 02, PP. 18-31, 2024

through machine learning approaches. The next section of this paper provides an outline of the emerging trends as
well as areas for future studies, which captivates readers with different perspectives on the Vietnamese AI methods
in medical image processing processes.
In [7], it would entail a literature review on the descriptive aspects of automating machine learning (AutoML), the
reasons for which it can assist healthcare custodians in allocating their appropriate effort, and the required
knowledge in the use of the machine learning models. With the objective of identifying future opportunities as
well as the drawbacks of using AutoML in healthcare, the present study has come up with future opportunities and
disadvantages associated with AutoML. The next part looks into the ways in which AutoML is used in the industry.
A study highlighted that automated options in AutoML could achieve performance levels that might be equal to
or even higher than the ones produced by humans when it comes to a certain kind of machine-learning task.
Efficiency by the unlimited nature of online learning is commonly within a shorter time, as proved by studying
101 scholarly articles. The advantage of AutoML is that it can be applied to a much larger scale than small- and
medium-sized retrospective datasets, allowing for enhanced data analysis, circuit design, or a variety of other tasks.
The need for healthcare professionals in machine learning is an underlying factor in the scant adoption of machine
learning techniques in hospitals, which is feeble in spite of its prospects, better health outcomes, low cost, and
advanced clinical research. One of the main barriers to be faced in this work is related to the computational
automation of model selection, model composition, and model parameterization. The objective is to increase the
use of machine learning techniques and distributed tasks among technology professionals. However, AutoML is
already being used for healthcare, but it is essential to pay more attention and deal with the emergence of the
common pattern of deployment. Breast cancer ranked as the second leading deadly disease of women, highlights
the seriousness of accurate early detection because of the effect it has on fatality rates. The implementation of
computer-aided detection provides radiologists with high precision in spotting abnormalities in medical pictures
and can be employed as vital data sources for identifying and diagnosing diseases. The presented study of the
different uses of deep learning and machine learning techniques in breast cancer, identification and classification
offers a holistic summary of the development in this area. [8] The study is about the application of multimodal
medical imaging methods to classify breast cancer. It describes the process of classifying the masses packed,
tumors, and non-tumors through various imaging techniques. The study will start by carrying out a rigorous
analysis of a variety of machine learning methods. This article reviews different deep learning models as well as
custom-made architectures that are specifically made for identifying and classifying breast cancer. In order to get
a complete picture, a summary of the imaging methods will be given.
[9] covers ML and AI utilization in different processes such as screening, predicting, forecasting, contact tracing,
and drug design in response to the pandemic caused by SARS-CoV-2 and other similar crises, with the pandemic
being a global issue where COVID-19 is a con. The studies on the contribution of machine learning and artificial
intelligence against the new coronavirus have been stretched by researchers who captured the successful
implementation of those technologies in past epidemics and disasters. This research undertakes a comprehensive
evaluation of works by scholars, finely reading through their abstracts, technologies, and findings in order to
determine the practicality of using machine learning (ML) and artificial intelligence (AI) to curb the crisis. The
dialogue takes us through the various study works that emphasize the intricate role of precision medicine and
discuss the already-made errors, as well as giving suggestions for researchers, model developers, medical
practitioners, and policymakers in the development of countermeasures for the existing COVID-19 and any
paramount challenges that may emerge as a result of the medical conditions. Contrarily, the AI and ML innovations
that are advancing with each passing day demonstrate great potential for halting further pandemics of the SARS-
CoV-2 virus and downscaling human intervention in healthcare operations. Over recent years, it has been observed
that data-driven techniques are taking on the appropriate leading role in some scientific domains, while
technological phrases like networked system technology and data collection are constantly expanding. In [10],
there is no doubt that the data revolution has strongly affected the life sciences, where the huge volume of data
will ultimately lead to the appearance of many new approaches to data management and analysis. The application
of machine learning and computational intelligence capabilities is particularly relevant to complicated situations
with a wealth of data. Even though the growing complexity of nonlinear models leads to concerns about their
interpretability and explainability, they still provide a promising pathway toward the discovery of new knowledge
in scientific fields. The difficulty in paying attention to this strife is crushed by the prominence of the sector in the
four front-line sectors like medicine and health care (which may lead to ineffective deployment of machine
learning solutions in dependency on machine learning mechanization). The overall objective of this study is to
examine the latest research work done in regard to machine learning techniques as far as their interpretability and
explainability in medicine and the healthcare industry in particular. This talk cements the case for the integration
of the methodology above for data and diagramming into the process of solving readability problems.

22
DOI: https://doi.org/10.54216/JAIM.070202
Received: May 17, 2023 Revised: September 19, 2023 Accepted: February 17, 2024
Journal of Artificial Intelligence and Metaheuristics (JAIM) Vol. 07, No. 02, PP. 18-31, 2024

Deep learning (DL) and machine learning (ML) technologies for computer-aided detection have been widely used
in the medical industry in the past few years. The importance of medical images to modern medicine lies in the
fact that they are the main source of diagnostic data. They possess an urgent responsibility in the timely detection
of diseases, being a critical factor in the reduction of the death rate due to diseases such as cancer and tumors.
Radiologists and physicians do this using several methods, such as an elaborate inspection that is carried out with
the aim of acquiring just the necessary characteristics. Machine learning has many limitations in terms of the
amount of data it can process. Still, deep learning expertise lies in the ability to handle any size of data smoothly.
Deep learning, also known as DL, is an improved form of machine learning that works with multilayered neural
networks to take in a wider variety of features from data sets. This study examines the application of ML and DL
algorithms to the detection and classification of various diseases, which entails a thorough literature review. A
total of 40 primary articles, which have been collected from reputable scholarly journals and conferences for the
last 8 years, were included in the present research. This study presents the exploration of some ML and DL
techniques that are mostly used in medical imaging, assessment methods and operations, and also patterns for
datasets. In addition, the sequence of experiments carried out in employing the MRI dataset to conduct a
comparative analysis of machine learning classifiers and deep learning models is also considered. The main goal
of [11] is to provide the healthcare community with the required tools. This means that the personnel will be able
to make wiser decisions about the diagnostic methods for some diseases and, as a result, achieve even higher
effectiveness and accuracy. The field of medicine has made serious progress in data collection and deep learning
algorithms; thus, artificial intelligence (AI) and machine learning (ML) have had a remarkable influence. Making
use of the peculiar hold of the orthopedics discipline, which, among other things, offers big data and the possibility
of employing big data, orthopedic surgeons can thus improve their level of treatment in different ways. [12]
presents a comprehensive textbook style, highlighting the latest scholarly work related to machine learning (ML)
applications in orthopedics.
The literature review concludes that the level of machine learning influence in the medical domain is expected to
grow and drive a paradigm shift that will cause medical care to be perceived in a new way. The findings of the
existing research illustrate how machine learning enhances diagnostics, prognosis, and treatment management in
healthcare. Indeed, therein lies the promise, but interpretability, ethics, and the accuracy of validation are still
being worked on. With ML applications still growing and advancing, a corresponding bridge between innovation
and practical implementation is, to date, a topical and most crucial aspect. The review shows how machine learning
can change medical practices, saying that the researchers, the practitioners, and the policymakers need to partner
to work through the dynamics of the evolving landscape in favor of the patient's care and the results.
3. Machine Learning
Since robots are not inherently intelligent, the objective of allowing them to develop learning abilities similar to
human learning might be considered aspirational [13-16]. Diversity is one of the ways that human and machine
labor differs from one another; intellect is another. Robots can't learn from their history, but people can. Giving
detailed production instructions is essential, and they have to be followed by others [17-19]. Machine learning
enables computers to learn from experience by allowing the construction of experience databases. For instance,
the majority of the regulations controlling traditional computer systems in the past were. The computer has relied
on these instructions to get it through every obstacle it has encountered over the years. Entities in the system: In
spite of this, computers may now make decisions based on rules that are produced by algorithms rather than by
human writers thanks to machine learning. This "soft coding" process is what causes memory loss. Machine
learning is one type of artificial intelligence. Intelligence: Machine learning (ML)-based gadgets are designed to
function without human assistance and to develop higher levels of intelligence. Machine learning makes machines
smarter and less dependent on human support. In actuality, the term "smart machine" is a symbol [20]. As a result,
the topic of discussion is machine learning and its goal. Allan Turing posed the question, "Is a machine eligible to
be involved in the cognitive function of human beings?" in 1995. This is how he came up with the "Turing test," a
method used by human interrogators to compare and assess people against robots. This quotation assesses the
cognitive capacities of the machine [21, 22]. These days, it's a popular misperception that machine learning
requires mental models. Arthur Samuel summarizes the area of machine learning as focusing on making computers
more capable of learning on their own without requiring explicit programming instructions [23].

23
DOI: https://doi.org/10.54216/JAIM.070202
Received: May 17, 2023 Revised: September 19, 2023 Accepted: February 17, 2024
Journal of Artificial Intelligence and Metaheuristics (JAIM) Vol. 07, No. 02, PP. 18-31, 2024

As was previously said, machine learning (ML) is considered a popular application of artificial intelligence (AI)
in which computers, software, and hardware all function via cognition, or something very close to how the human
brain functions. These days, almost every industry has at least one machine learning approach that we utilize on a
daily basis without even recognizing it, such as email spam and virus screening. Additional examples of ML
applications are shown in Figure 1. This review discusses the application of machine learning (ML) to image
processing, diagnostics, and healthcare.

Figure 1: Machine Learning Applications

The steps required in developing a learning model for the healthcare sector are described in this section. Remember
that this section aims to provide researchers with an understanding of the process of developing a medical learning
model. We recommend that scholars evaluate and carry out further research in this area to have a comprehensive
understanding of and knowledge of learning models. When developing a learning model for the healthcare sector,
there are five primary steps that must be considered: issue description, dataset, data preparation, ML model
construction, and assessment. Figure 2 illustrates the steps. Below is a thorough explanation of each of these stages.

Figure 2: Framework for Designing a Learning Model

3.1. Machine Learning Methods in Healthcare
An ML-based technology as a whole has become imperative for CAD, changing the characteristics of disease
determination and prognosis. Taught using large sets of samples that were obtained from diagnostic tests' outcomes
and experts' diagnoses, they computerized medical examinations, which helps clinicians to get highly dependable
and trustable results within a short period and also make predictions. Via the ML techniques, diagnostic systems
can be endowed with state-of-the-art recognition and calibration, bio-markers are being found earlier, and
diagnosis is right on target. Wei et al. demonstrated the potential of machine learning in breast cancer diagnosis
through their innovative approach [24-26]. They developed machine learning models to automatically identify
clustered microcalcifications in digital breast mammography, a crucial aspect of breast cancer diagnosis. Their
system, trained on a vast database of 697 mammograms, enabled radiologists to distinguish between benign and
malignant clusters of microcalcifications with high accuracy, thereby enhancing the efficiency and precision of
24
DOI: https://doi.org/10.54216/JAIM.070202
Received: May 17, 2023 Revised: September 19, 2023 Accepted: February 17, 2024
Journal of Artificial Intelligence and Metaheuristics (JAIM) Vol. 07, No. 02, PP. 18-31, 2024

breast cancer diagnosis. Shoeb and Guttag's introduction of an SVM patient-specific classifier using EEG scalp
data is a testament to the life-saving potential of machine learning in healthcare. Despite the complex dynamics of
brain electrical signals, their approach successfully classified electroencephalographic patterns related to seizures
from those related to other brain activities. This early detection of seizures, before the onset of serious symptoms,
holds the promise of saving lives and underscores the transformative impact of ML in healthcare [27].
In the study of Ye et al., a supervised machine learning approach called compound covariate predictor was
implemented to classify patients with metastatic hepatocellular carcinoma (HCC). Stunning results were reported,
and significant genetic markers emerged that were found to influence survival and metastasis [28]. Using the
random combination of 50 HCC test samples that were collected from 30 patients, they were able to achieve high
accuracies; thus, interpreting individualized treatment options portends promising results. In addition, it is stated
that the mentioned algorithm, an artificial intelligence (AI) based on supervised feature projection, is used as a
diagnostic tool for cardiac arrhythmias with the help of a bigger dataset consisting of 452 patients with 279
measurable characteristics, including the clinical features and ECG signals [29]. Their algorithm attained this result
through rigorous training on pre-classified samples, and following butcher rigid cross-validation, it has shown
robust performance, being on one level with the most used classification algorithms in terms of accuracy and
effectiveness.
Moreover, in the study by Wu and others, they undertook pioneering research to build a predictive model for heart
failure where they were able to accurately detect a case of heart failure at least six months before the appearance
of the symptoms. Using data from 536 patient files obtained between 2001 and 2006 in their study, the team
employed three different machine learning methods: boosting, support vector machines (SVM) and logistic
regression. These were all tested through extensive 10-fold cross-validation analyses, which validate their accuracy
in picking heart failure before it occurs [30]. While incorporating artificial intelligence and machine learning into
medicine may seem like a new way of doing things, these pioneering endeavors clearly highlight the transformative
capacity of these technologies in improving diagnosis, prediction and treatment outcomes when applied to
medicine.
3.2. Machine Learning in Medical Applications
ML is frequently utilized to solve problems in healthcare and medicine [31]. The next sections will cover a few
areas where CSA was applied: wearable sensors, brain problems, medical imaging, cancer (types and diagnostics),
and so on. This section's classifications (i.e., problem-based data presentation and, in some situations, machine
learning techniques-based data display) are derived from the contributions of previous research [32].
3.2.1. Prediction of Cancer
Cancer research is a significant area of study with broad societal implications. A variety of areas in which the use
of machine learning (ML) to cancer research shows tremendous potential include benchmarking cancer-related
issues including medication response, therapeutic options, and the categorization and prediction of cancer types.
Shi et al. [33] presented a hybrid machine learning approach that combines SVM with the k-nearest neighbors
(KNN) and k-top-scoring pair (TSP) feature ranking algorithms. The proposed method was motivated by the
capacity of feature selection approaches to improve the gene analysis-based categorization and prediction of
complicated illnesses. The hybrid technique, which has shown high accuracy performance, is validated using four
actual and simulated cancer prognostic datasets. Furthermore, ML models may be used to predict which drugs will
function effectively in a clinical setting, as well as the sorts of therapies that will be most effective for a particular
patient. Consequently, Menden et al. developed artificial neural network-based machine learning models to predict
the response of cancer cell lines to different types of treatment, as shown by IC50 values. The prediction technique
is based on the genetic and chemical properties of the drug's cell lines. The genetic background of each cell was
considered for calculating the IC50 profile. They found that depending on how the chemicals form, hundreds of
pharmaceuticals may have their potential efficacy as anti-tumor agents assessed in silico. This result may be helpful
for customized therapy and lead to the discovery of new potential for medication repositioning by connecting the
sensitivity of the drug to the patient's genetic makeup. The clinical efficacy of cancer medications is predicted
using the same three machine learning (ML) methods: SVM, BT, and RF. These techniques transfer features
derived from expression-based cell line data. Additionally, a novel approach to drug grading and customized
medicine was given by Borisov et al. [34]. The algorithms were tested on many datasets of cancer-like illnesses,
including chronic myeloid leukemia, lung adenocarcinoma, and kidney carcinoma.
However, Huang et al. introduced an open-source software platform that predicts a tailored drug's responses based
on gene expression patterns by combining the SVM algorithm with the recursive feature elimination (RFE) model
[35]. Particular models were built using data on drug response and gene expression from the NCI-60 panel of cell
25
DOI: https://doi.org/10.54216/JAIM.070202
Received: May 17, 2023 Revised: September 19, 2023 Accepted: February 17, 2024
Journal of Artificial Intelligence and Metaheuristics (JAIM) Vol. 07, No. 02, PP. 18-31, 2024

lines. The models accurately predict the drug receptiveness of several cancer cell lines. When the models were
used, predicted outcomes were obtained that matched previously identified answers from the literature. It is
expected that several cancer types will be used to evaluate the proposed open-source software platform. A critical
first step in applying machine learning to cancer research is selecting the appropriate software to utilize to mitigate
cancer-related problems. Wozniak et al. suggested a CANDLE/Supervisor system to deal with hyper-parameter
exploration in deep neural networks. Many elements are provided by the proposed framework to facilitate machine
learning in cancer research. Users have the option to modify the ML problem or optimizer at first. It also allows
the use of datasets with different scales. Finally, it will enable users to access a range of computational
concurrencies across several leadership-class systems. When the program was evaluated on individual
workstations and clusters, good results were observed for scalability and multiplatform execution [36].
The Cancer's types:
1. Breast cancer

Breast cancer is the most frequent cancer globally, and it is also the leading cause of cancer-related mortality for
women. It starts to develop within the breast tissue. Conversely, timely identification and precise diagnosis might
contribute to a higher chance of survival [37]. Breast cancer is the second most common cause of cancer-related
death in women (BDC). It is also one of the most temporary types of cancer if caught early. The authors have
attempted to report that SVM has a high number of experiences with accurate diagnosis with ever higher levels of
confirmation. It tries to support the accurate analytical power of SVM [38]. This work proposed to assess breast
cancer by combining SVM with a feature selection approach. The research made use of a number of popular
datasets, including the Wisconsin Breast Cancer Database (WBCD). Machine learning techniques for the study of
breast cancer often make use of these datasets. Specificity, sensitivity, classification performance, positive and
negative predictive values, receiver running characteristic confusion matrix, and curves are used to evaluate the
efficacy of the proposed method. The findings showed that the SVM model with five properly selected features
produced a maximum classification accuracy of 99.5%. This algorithm appears to be more promising than the one
that was previously disclosed.
2. Lung cancer

Regarding the number of new cases (about 0.13 annually) and deaths (around 0.2 annually), lung cancer remains
one of the most well-known malignancies globally. Inaccuracies in the lung cancer sample or poor growth
judgment led to unsuccessful surgeries since the anticancer approach depends on the tumor form. The effectiveness
of machine learning algorithms in the study and categorization of lung cancer based on gene expression stages was
assessed in [39]. The University of Michigan, the University of Toronto, the Dana-Farber Cancer Institute, and
Brigham and Women's Hospital were the four publicly accessible datasets that were analyzed in that work. There
were 203, 96, 39, and 181 units in each of the databases. The k-nearest neighbor approach, naive Bayes, and
decision trees were used to distribute the data, along with the assumption of a few normal attribute distributions.
When ML approaches were tested for effectiveness, SVM yielded the best results across all datasets. All the
algorithms—except for the decision tree—showed ideal performance. However, the decision tree produced the
best outcomes for the datasets from the University of Toronto. All in all, lung cancer morphology analysis and
associated activities can benefit from the application of machine learning techniques based on gene expression
level validation. The problem of utilizing protein mass colors to differentiate between 24 ill and 15 healthy people
was discussed in [40]. Baseline removal, volume of charge ratio (m/z) normalization, and the conversion of certain
peak height ratios to top ratios were among the preprocessing steps performed on the data. After preprocessing the
data, the biggest challenge was the enormous number of variables (1676 m/z values) compared to the number of
instances. The dimension reduction approach was necessary for the analysis. Model creation was linked to choice
selection in a solitary 10-fold cross-validation cycle. The 1676-mass and 124-mass datasets were used to explore
a range of experimental configurations, including two peak top representations, pair choice selection approaches,
and six selection procedures. NB with a multilayer perceptron showed significant prediction accuracy, with NB
exhibiting harmonious efficacy in a range of experimental setups. They attempted to determine which proteins
were the most differentiable by utilizing NN-based sensitivity evaluation and rates from the pair variable selection
approaches. These three designs determined which four peaks—11683, 1403, 17350, and 66107—were the more
important discriminators. A large data method for clinical decision-making to identify pneumonitis after
stereotactic frame emission therapy (SBRT). Of the 201 cases of SBRT that followed, eight cases (4.0%) had
broadcast pneumonitis (RP) among the 61 features that were documented. Using decision projections,
pneumoniaitis ratios for every characteristic were ascertained independently. The efficacy of three distinct
algorithms was compared. The three distinct fundamental characteristics chosen for multivariate analysis were the
dosage of the heart (15), the bronchus (4), and the culture. A higher level of accuracy was achieved if the
26
DOI: https://doi.org/10.54216/JAIM.070202
Received: May 17, 2023 Revised: September 19, 2023 Accepted: February 17, 2024
Journal of Artificial Intelligence and Metaheuristics (JAIM) Vol. 07, No. 02, PP. 18-31, 2024

recommended algorithm was applied consistently. It was estimated that to identify the distribution of pneumonitis
with an error of less than 10%, an individual size of 800 cases would be needed. The quantity of instances in the
inquiry restricted the accuracy of the classification, not the views acquired.
4. Prostate cancer

Over the past few years, there has been significant growth in the use of computers for cancer diagnosis. Machine
learning is one technique widely used to detect and predict prostate cancer. This section covers the methods for
improving prostate cancer diagnosis with machine learning. Zhu et al. proposed an online adaptive radiotherapy
tool to evaluate adaptive intensity-modulated radiation treatment (IMRT) using machine learning (ML)
approaches—their proposed method aimed to ascertain the prostate adaptive IMRT quality threshold and
prerequisites [41]. Another ML-based prostate cancer diagnostic technique was presented by Hussain et al. [42].
Prostate cancer was successfully diagnosed using SVM and Bayesian multi-ML approaches. A couple of feature
extraction methods were also used for even more efficiency gains. To enhance PI-RADS v2's performance,
proposed an analysis for magnetic resonance radiomics. Evaluating whether machine learning approaches may
improve prostate cancer diagnosis was the primary objective of the experiment.
5. Brain cancer

The human brain is the most complex organ, with around 100 billion neurons and one million billion (1015)
connections [43]. It controls complex functions like emotions and feelings, as well as sensorimotor tasks like
breathing and walking, as well as cognitive functions like speaking, thinking, and memorizing. Furthermore, the
brain can develop a variety of conditions that require surgery, result in long-term impairment, or even deteriorate
the functions mentioned above. Consequently, functional magnetic resonance imaging (fMRI) has emerged as a
powerful new modality for obtaining massive volumes of data on brain activity. The focus of Zacharaki et al.'s
research [44] was on the texture and form of MRI images by categorizing brain tumor kind and grade. The
automated computer analysis tool outperformed human readers, improving the consistency and accuracy of
procedures for diagnosing brain tumors. Thus, a computer-assisted classification strategy integrating perfusion and
conventional MRI was devised for differential diagnosis. After using feature extraction of the tumor form and
intensity characteristics, the authors employed SVM feature selection with recursive feature removal. A non-
invasive machine learning tool is EEG. This kind of electrophysiological monitoring captures the electrical activity
in the brain. Consequently, focused on biometrics to understand and distinguish persons based on their behavioral
characteristics After developing the Visual Evoked Potential-based biometrics framework, every result passed the
feature extraction and classification stages. A brief study has been conducted on the efficacy of preprocessing and
classification strategies for brain-computer interface (BCI) and mental state monitoring via electroencephalogram
(EEG). BCI makes it possible to communicate only through brain impulses, independent of muscles or peripheral
nerves. An overview of the Berlin brain-computer interface (BBCI), which is thought to be a non-invasive
technique, was also provided by the writers. The findings are presented from an experiment that included real-time
arousal monitoring.
3.2.2. Medical Imaging
The following discussions in this section are also supported by the principles of fMRI, MRI, and EEG, which are
the foundation of radiological imaging. As one of the first doctors to have access to such modes of imaging, the
radiologists are proficient and expert at reading the radiologic images, using the ML technique intensely to increase
and strengthen their knowledge and ability to make a perfect diagnosis. Radiography remains a basic but modern
tool at the disposal of medical professionals, now comprised of CT, MRI, PET, and tomographies, which form the
pillars of modern medical imaging. For example, Zhang et al. demonstrated the achievement of the first sequence
enhancement phase learning to the specific SVM machine, which is able to increase discriminatory ability only
for the identification of the microcalcifications (MC) in digital mammography, thus can significantly improve the
diagnostic precision and efficiency [45].
Alongside this, their tumor segmentation method, which uses an SVM model to swap over the intricate, non-linear
distribution of data from the tumor, assuming there is no prior information, represents a pivotal step into on-
imaging. In contrast, El-Naqa et al. focus on the utilization of a specific technology – SVM, to effectively reduce
classification error, which is the main barrier to the accurate detection of microcalcifications; therefore, SVM is
known as the way to convert classification tasks into supervised learning ones. Analysis of SVM, together with
microcalcifications at every image position, is used to improve diagnostic workflow, thus helping to shift from the
old X-ray practice to the new one. The link at the intersection of machine learning and radiology shows a new
phase in medical imaging, which not only provides a deeper peek into the biological mechanisms but also
concurrently improves diagnostic accuracy and performance and enriches patients' outcomes. Nevertheless,
27
DOI: https://doi.org/10.54216/JAIM.070202
Received: May 17, 2023 Revised: September 19, 2023 Accepted: February 17, 2024
Journal of Artificial Intelligence and Metaheuristics (JAIM) Vol. 07, No. 02, PP. 18-31, 2024

machine learning (ML) techniques are besidered in the processes in ways that exclude the public and feed the well-
being of disadvantaged groups. As pointed out by Nyako et al., it is the in-depth exploration of learning
mechanisms such as Lemm et al.'s comprehensive review of adoption processes in the field and its subsequent
contribution towards a universal understanding of the incorporation of these methodologies from both theoretical
and practical perspectives thereby enabling access to the field by all.
In addition to this, the particular projects that highlighted the wide range of uses of radiography and AI technology
in the past six domains showed the enormous transforming capability that radiographic imaging and machine
learning have in determining diagnosis computer-aided detection and organ segmentation and registration [46-47].
This harmonization may water this; not only can we expect diagnostic accuracy and efficiency to be revolutionized,
but patient care may progress through cost-effectiveness and expanded clinical expertise. While the two fields
progress together, they will jointly pave the way for more thorough diagnoses as well as newer and more innovative
treatment procedures that will ultimately usher in the dawn of a more profitable and rewarding healthcare delivery
system.
5. Conclusion
This paper has rigorously and comprehensively explored the use of machine learning methods in medical
diagnostics, delving deep into the newest findings and trends that have encountered and shaped this dynamic field.
Rounding these more than fifty scholarly articles are the meticulous details of the intricate connection between
machine learning techniques and the medical challenges, which we are going to expound on with its specifics.
Focusing on machine learning applications in the multidimensional health diagnostics arena, this research piece
endeavors to precisely investigate the plethora of benefits, disadvantages, and strengths in employing machine
learning algorithms in various health sector cases. This study highlighted various layers of complexities, and, in
response to that, a series of the most compelling lines for exploring further and conducting scientific exercises
were presented. First and most importantly, the application of supervised machine learning methods of cancer
detection models highlights vital aspects, including its capability and imperfections. The supervised ML, however,
in many cases, manages to assign future inputs to predefined classes properly. However, there are two main
challenges with these methods: either class labels will be wrong, and the algorithm will only be consistent for
some datasets. The idea of implementation progression strategies, this study demands a more detailed touch to halt
and conquer these challenges effectively and improve the cancer detection algorithm's performance. In addition,
the paper leads us to understand neural diagnosis as machine learning might hold revolutionizing diagnosis of
conditions like epilepsies and trauma. Realizing the complementary advantages of the merging of artificial
intelligence and machine learning capabilities, the vision powers the enhancement of ML effectiveness via robust
monitoring of critical interest zone diagnostics, with the goal of improving diagnosis precision and efficiency.
On the other hand, the diversity of classifiers is imparted by incorporating fuzzy classifiers in diagnostics, which
have not been exploited before. This suggestion does so by giving priority to methods and technologies that would
enhance the existing medical practitioner and researcher's capabilities. This clearly manifested the need to find
new approaches to tackle the complicated challenges in medical diagnostics. Furthermore, the research pushes
researchers to aspire to use unsupervised learning approaches, which stand out as top-notch intelligence that can
hold and learn large amounts of medical image data. Leveraging the intrinsic features of unsupervised learning as
an integral part of this recommendation underscores the amazing power of this capability, which can be effectively
used to scale the current conventional medical imaging landscape to never-before-seen levels of efficiency with
exceptional output. Summarily, the survey's overall objective is to be an information and guiding resource for
researchers, stars of the industry, and decision-makers as a whole and to set off an irrevocable cycle of innovation
in the medical diagnostics field. In an attempt to scaffold the most important outcomes of the advanced studies,
our research tries to give a direction to future studies, which will lead to tough challenges and eventually produce
therapeutic achievements to be used in the medical field. As for future directions, this project shall pursue the
creation of an interactive website that is a dynamic information resource, which will allow for collaboration,
innovation, and dissemination of the most recent medical news and insights not only in the existing field but also
in the development.
References
[1] Emmert-Streib, F. (2021). From the Digital Data Revolution toward a Digital Society: Pervasiveness of
Artificial Intelligence. Machine Learning and Knowledge Extraction, 3(1), Article 1.
https://doi.org/10.3390/make3010014
[2] Gruetzemacher, R., & Whittlestone, J. (2022). The transformative potential of artificial intelligence.
Futures, 135, 102884. https://doi.org/10.1016/j.futures.2021.102884

28
DOI: https://doi.org/10.54216/JAIM.070202
Received: May 17, 2023 Revised: September 19, 2023 Accepted: February 17, 2024
Journal of Artificial Intelligence and Metaheuristics (JAIM) Vol. 07, No. 02, PP. 18-31, 2024

[3] Ramezani, M., Takian, A., Bakhtiari, A., Rabiee, H. R., Fazaeli, A. A., & Sazgarnejad, S. (2023). The
application of artificial intelligence in health financing: A scoping review. Cost Effectiveness and
Resource Allocation : C/E, 21, 83. https://doi.org/10.1186/s12962-023-00492-2
[4] Ma, L., & Sun, B. (2020). Machine learning and AI in marketing – Connecting computing power to
human insights. International Journal of Research in Marketing, 37(3), 481–504.
https://doi.org/10.1016/j.ijresmar.2020.04.005
[5] El-kenawy, E.-S. M., Albalawi, F., Ward, S. A., Ghoneim, S. S. M., Eid, M. M., Abdelhamid, A. A., Bailek,
N., & Ibrahim, A. (2022). Feature Selection and Classification of Transformer Faults Based on Novel
Meta-Heuristic Algorithm. Mathematics, 10(17), Article 17. https://doi.org/10.3390/math10173144
[6] El-kenawy, E.-S., Ibrahim, A., Mirjalili, S., Eid, M., & Hussein, S. (2020). Novel Feature Selection and
Voting Classifier Algorithms for COVID-19 Classification in CT Images. IEEE Access, 8, 179317–
179335. https://doi.org/10.1109/ACCESS.2020.3028012
[7] Abdelhamid, A. A., El-Kenawy, E.-S. M., Ibrahim, A., Eid, M. M., Khafaga, D. S., Alhussan, A. A.,
Mirjalili, S., Khodadadi, N., Lim, W. H., & Shams, M. Y. (2023). Innovative Feature Selection Method
Based on Hybrid Sine Cosine and Dipper Throated Optimization Algorithms. IEEE Access, 11, 79750–
79776. https://doi.org/10.1109/ACCESS.2023.3298955
[8] Deng, W., Chen, R., He, B., Liu, Y., Yin, L., & Guo, J. (2012). A novel two-stage hybrid swarm
intelligence optimization algorithm and application. Soft Computing, 16(10), 1707–1722.
https://doi.org/10.1007/s00500-012-0855-z
[9] Bolón-Canedo, V., Sánchez-Maroño, N., & Alonso-Betanzos, A. (2016). Feature selection for high-
dimensional data. Progress in Artificial Intelligence, 5(2), 65–75. https://doi.org/10.1007/s13748-015-
0080-y
[10] Molina, L. C., Belanche, L., & Nebot, A. (2002). Feature selection algorithms: A survey and experimental
evaluation. 2002 IEEE International Conference on Data Mining, 2002. Proceedings., 306–313.
https://doi.org/10.1109/ICDM.2002.1183917
[11] El-kenawy, E.-S. M., Khodadadi, N., Mirjalili, S., Abdelhamid, A. A., Eid, M. M., & Ibrahim, A. (2024).
Greylag Goose Optimization: Nature-inspired optimization algorithm. Expert Systems with Applications,
238, 122147. https://doi.org/10.1016/j.eswa.2023.122147
[12] Dorigo, M., Birattari, M., & Stutzle, T. (2006). Ant colony optimization. IEEE Computational Intelligence
Magazine, 1(4), 28–39. https://doi.org/10.1109/MCI.2006.329691
[13] El-Kenawy, E.-S. M., Eid, M. M., Abdelhamid, A. A., Ibrahim, A., Takieldeen, A. E., & Elkhalik, S. H.
A. (2022). Hybrid Particle Swarm and Gray Wolf optimization for Prediction of Appliances in Low-
Energy Houses. 2022 International Telecommunications Conference (ITC-Egypt), 1–5.
https://doi.org/10.1109/ITC-Egypt55520.2022.9855690
[14] Papa, J. P., Rosa, G. H., de Souza, A. N., & Afonso, L. C. S. (2018). Feature selection through binary
brain storm optimization. Computers & Electrical Engineering, 72, 468–481.
https://doi.org/10.1016/j.compeleceng.2018.10.013
[15] Eid, M. M., El-kenawy, E.-S. M., & Ibrahim, A. (2021). A binary Sine Cosine-Modified Whale
Optimization Algorithm for Feature Selection. 2021 National Computing Colleges Conference (NCCC),
1–6. https://doi.org/10.1109/NCCC49330.2021.9428794
[16] Arora, S., & Anand, P. (2019). Binary butterfly optimization approaches for feature selection. Expert
Systems with Applications, 116, 147–160. https://doi.org/10.1016/j.eswa.2018.08.051
[17] Chuang, L.-Y., Tsai, S.-W., & Yang, C.-H. (2011). Improved binary particle swarm optimization using
catfish effect for feature selection. Expert Systems with Applications, 38(10), 12699–12707.
https://doi.org/10.1016/j.eswa.2011.04.057
[18] Salimi, H. (2015). Stochastic Fractal Search: A powerful metaheuristic algorithm. Knowledge-Based
Systems, 75, 1–18. https://doi.org/10.1016/j.knosys.2014.07.025
[19] Abdelhamid, A. A., El-Kenawy, E.-S. M., Ibrahim, A., Eid, M. M., Khafaga, D. S., Alhussan, A. A.,
Mirjalili, S., Khodadadi, N., Lim, W. H., & Shams, M. Y. (2023). Innovative Feature Selection Method
Based on Hybrid Sine Cosine and Dipper Throated Optimization Algorithms. IEEE Access, 11, 79750–
79776. https://doi.org/10.1109/ACCESS.2023.3298955
[20] Abdollahzadeh, B., & Gharehchopogh, F. S. (2022). A multi-objective optimization algorithm for feature
selection problems. Engineering with Computers, 38(3), 1845–1863. https://doi.org/10.1007/s00366-
021-01369-9
[21] Aghdam, M. H., Ghasem-Aghaee, N., & Basiri, M. E. (2009). Text feature selection using ant colony
optimization. Expert Systems with Applications, 36(3, Part 2), 6843–6853.
https://doi.org/10.1016/j.eswa.2008.08.022

29
DOI: https://doi.org/10.54216/JAIM.070202
Received: May 17, 2023 Revised: September 19, 2023 Accepted: February 17, 2024
Journal of Artificial Intelligence and Metaheuristics (JAIM) Vol. 07, No. 02, PP. 18-31, 2024

[22] Al-Tashi, Q., Abdul Kadir, S. J., Rais, H. M., Mirjalili, S., & Alhussian, H. (2019). Binary Optimization
Using Hybrid Grey Wolf Optimization for Feature Selection. IEEE Access, 7, 39496–39508.
https://doi.org/10.1109/ACCESS.2019.2906757
[23] Arora, S., & Anand, P. (2019). Binary butterfly optimization approaches for feature selection. Expert
Systems with Applications, 116, 147–160. https://doi.org/10.1016/j.eswa.2018.08.051
[24] Chuang, L.-Y., Tsai, S.-W., & Yang, C.-H. (2011). Improved binary particle swarm optimization using
catfish effect for feature selection. Expert Systems with Applications, 38(10), 12699–12707.
https://doi.org/10.1016/j.eswa.2011.04.057
[25] Dwivedi, R., Tiwari, A., Bharill, N., & Ratnaparkhe, M. (2023). A Novel Clustering-Based Hybrid
Feature Selection Approach Using Ant Colony Optimization. Arabian Journal for Science and
Engineering, 48(8), 10727–10744. https://doi.org/10.1007/s13369-023-07719-7
[26] El-Kenawy, E.-S. M., Eid, M. M., Saber, M., & Ibrahim, A. (2020). MbGWO-SFS: Modified Binary Grey
Wolf Optimizer Based on Stochastic Fractal Search for Feature Selection. IEEE Access, 8, 107635–
107649. https://doi.org/10.1109/ACCESS.2020.3001151
[27] Emary, E., Zawbaa, H. M., & Hassanien, A. E. (2016). Binary grey wolf optimization approaches for
feature selection. Neurocomputing, 172, 371–381. https://doi.org/10.1016/j.neucom.2015.06.083
[28] Ghaemi, M., & Feizi-Derakhshi, M.-R. (2016). Feature selection using Forest Optimization Algorithm.
Pattern Recognition, 60, 121–129. https://doi.org/10.1016/j.patcog.2016.05.012
[29] Kamyab, S., & Eftekhari, M. (2016). Feature selection using multimodal optimization techniques.
Neurocomputing, 171, 586–597. https://doi.org/10.1016/j.neucom.2015.06.068
[30] Khodadadi, N., Khodadadi, E., Al-Tashi, Q., El-Kenawy, E.-S. M., Abualigah, L., Abdulkadir, S. J.,
Alqushaibi, A., & Mirjalili, S. (2023). BAOA: Binary Arithmetic Optimization Algorithm With K-Nearest
Neighbor Classifier for Feature Selection. IEEE Access, 11, 94094–94115.
https://doi.org/10.1109/ACCESS.2023.3310429
[31] Liu, Y., Wang, G., Chen, H., Dong, H., Zhu, X., & Wang, S. (2011). An Improved Particle Swarm
Optimization for Feature Selection. Journal of Bionic Engineering, 8(2), 191–200.
https://doi.org/10.1016/S1672-6529(11)60020-6
[32] Mafarja, M., & Mirjalili, S. (2018). Whale optimization approaches for wrapper feature selection. Applied
Soft Computing, 62, 441–453. https://doi.org/10.1016/j.asoc.2017.11.006
[33] Papa, J. P., Rosa, G. H., de Souza, A. N., & Afonso, L. C. S. (2018). Feature selection through binary
brain storm optimization. Computers & Electrical Engineering, 72, 468–481.
https://doi.org/10.1016/j.compeleceng.2018.10.013
[34] Song, X., Zhang, Y., Gong, D., & Sun, X. (2021). Feature selection using bare-bones particle swarm
optimization with mutual information. Pattern Recognition, 112, 107804.
https://doi.org/10.1016/j.patcog.2020.107804
[35] Tabakhi, S., Moradi, P., & Akhlaghian, F. (2014). An unsupervised feature selection algorithm based on
ant colony optimization. Engineering Applications of Artificial Intelligence, 32, 112–123.
https://doi.org/10.1016/j.engappai.2014.03.007
[36] Zawbaa, H. M., Emary, E., Parv, B., & Sharawi, M. (2016). Feature selection approach based on moth-
flame optimization algorithm. 2016 IEEE Congress on Evolutionary Computation (CEC), 4612–4617.
https://doi.org/10.1109/CEC.2016.7744378
[37] Khafaga, D., El-kenawy, E.-S., Alrowais, F., Kumar, S., Ibrahim, A., & Abdelhamid, A. (2022). Novel
Optimized Feature Selection Using Metaheuristics Applied to Physical Benchmark Datasets.
Computers, Materials and Continua, 74, 4027–4041. https://doi.org/10.32604/cmc.2023.033039
[38] Stańczyk, U. (2015). Feature Evaluation by Filter, Wrapper, and Embedded Approaches. In U. Stańczyk
& L. C. Jain (Eds.), Feature Selection for Data and Pattern Recognition (pp. 29–44). Springer.
https://doi.org/10.1007/978-3-662-45620-0_3
[39] Chen, G., & Chen, J. (2015). A novel wrapper method for feature selection and its applications.
Neurocomputing, 159, 219–226. https://doi.org/10.1016/j.neucom.2015.01.070
[40] Sánchez-Maroño, N., Alonso-Betanzos, A., & Tombilla-Sanromán, M. (2007). Filter Methods for Feature
Selection – A Comparative Study. In H. Yin, P. Tino, E. Corchado, W. Byrne, & X. Yao (Eds.), Intelligent
Data Engineering and Automated Learning—IDEAL 2007 (pp. 178–187). Springer.
https://doi.org/10.1007/978-3-540-77226-2_19
[41] Wang, S., Tang, J., & Liu, H. (2015). Embedded Unsupervised Feature Selection. Proceedings of the
AAAI Conference on Artificial Intelligence, 29(1), Article 1. https://doi.org/10.1609/aaai.v29i1.9211
[42] Dehghani, M., Trojovská, E., & Trojovský, P. (2022). A new human-based metaheuristic algorithm for
solving optimization problems on the base of simulation of driving training process. Scientific Reports,
12(1), Article 1. https://doi.org/10.1038/s41598-022-14225-7

30
DOI: https://doi.org/10.54216/JAIM.070202
Received: May 17, 2023 Revised: September 19, 2023 Accepted: February 17, 2024
Journal of Artificial Intelligence and Metaheuristics (JAIM) Vol. 07, No. 02, PP. 18-31, 2024

[43] Nematollahi, A. F., Rahiminejad, A., & Vahidi, B. (2017). A novel physical based meta-heuristic
optimization method known as Lightning Attachment Procedure Optimization. Applied Soft Computing,
59, 596–621. https://doi.org/10.1016/j.asoc.2017.06.033
[44] van der Schelling, M. (2021). A data-driven heuristic decision strategy for data-scarce optimization: With
an application towards bio-based composites.
https://repository.tudelft.nl/islandora/object/uuid%3Ad58271d6-21bb-470c-a5ee-4584b3b8ee29
[45] Fatih Güven, A., & Mahmoud Samy, M. (2022). Performance analysis of autonomous green energy
system based on multi and hybrid metaheuristic optimization approaches. Energy Conversion and
Management, 269, 116058. https://doi.org/10.1016/j.enconman.2022.116058
[46] Taghian, S., & Nadimi-Shahraki, M. H. (2019). A Binary Metaheuristic Algorithm for Wrapper Feature
Selection (Vol. 8, p. 172). https://doi.org/10.13140/RG.2.2.34937.90722
[47] Sahin, O., & Akay, B. (2016). Comparisons of metaheuristic algorithms and fitness functions on
software test data generation. Applied Soft Computing, 49, 1202–1214.
https://doi.org/10.1016/j.asoc.2016.09.045

31
DOI: https://doi.org/10.54216/JAIM.070202
Received: May 17, 2023 Revised: September 19, 2023 Accepted: February 17, 2024