IEEE - 56998

Monkeypox Prediction using Machine Learning

2023 14th International Conference on Computing Communication and Networking Technologies (ICCCNT) | 979-8-3503-3509-5/23/$31.00 ©2023 IEEE | DOI: 10.1109/ICCCNT56998.2023.10306370

Malti Bansal*, Riyanshi Arora, Sai Keshari, Sakshi Panchal

Department of Electronics & Communication Engineering,
Delhi Technological University (DTU), Delhi- 110042, India
*maltibansal@gmail.com

Abstract— The monkeypox virus is gradually reemerging as confirmed detecting monkeypox cases [1]. There are
COVID-19 viral infections decline globally. The recent currently 75 nations outside of Africa where there
outbreak in the early 2022 had grown fear and concern in are verified occurrences of monkeypox, making it a
people, making them believe that it would spread like
COVID-19. Public health organizations struggled to contain
significant global health concern. The Zoonotic
the present outbreak as healthcare professionals from all over Orthopox virus is responsible for the monkeypox
the world tried to become familiar with the many clinical signs infection. It belongs to the 'Poxviridae' type family
and treatment of this infection. Early identification is, and has strong ties to both cowpox and smallpox.
therefore, a crucial step to stop them from spreading broadly Although rodents and monkeys are still the main
throughout society. AI-based detection may be able to locate transmitters, cultural infection is still rather
them at an early stage. Such detection and prediction prevalent. In specific parts of Africa, which houses
techniques have brought about optimistic outcomes. It also
implies that the proposed approach is appropriate for mass
the majority proportion of tropical rainforests, a
screening by health practitioners. In image processing-based disease like monkeypox often affects a huge number
diagnostics, such as detection of cancel, identification of of individuals. The transmission is primarily caused
tumor cell, and COVID-19 patient detection, machine learning when the individual comes into a close contact (be it
has lately demonstrated great promise. Therefore, as physical or airborne through mucus, drops from
monkeypox spreads to human skin, a similar technique might respiratory tract) of any other animal, object or
be used to detect it. To help with the condition's subsequent human counterpart which is said to infected in the
diagnosis, a picture can be obtained and a dataset containing
similar images of the infection can be utilized to develop a
first place.
near-exact prediction model. In this review, we discuss the
clinical diagnosis, management, and improvement of the
classification accuracy of monkeypox images for development
of prediction models using ML. We also discuss the historical
and contemporary outbreaks of the monkeypox virus. This
study is being conducted in light of the ongoing epidemics in
various parts of the world.

Keywords — deep learning, detection, image processing,

machine learning, monkeypox, orthopoxvirus, smallpox,
VGG16

INTRODUCTION
I.
The emergence of the monkeypox virus as a disease Fig. 1: Monkeypox skin lesion through multiple phases
(Reproduced from [29])
in the year 2022, which was turned up by numerous
countries, revealed a new threat to the world at a
It is evident that artificial intelligence (AI) along
time when COVID-19 had already begun to have an
with Machine learning (ML); its sub-domain, is
impact. In 1990, around Central and West Africa,
growing in popularity across a wide spectrum of
there were mere 50 cases of the monkeypox
disciplines. With the special capabilities of ML,
infection. Whereas this number had increased to as
medical professionals can receive quick, accurate,
many as 5000 cases until the year 2020. Previously,
and safe imaging solutions. These solutions have
it was believed that monkeypox mostly affected
become widely recognized as useful decision-
Africa. In 2022, even so, a large number of non-
making aids. In 2019, different segmentation
African countries in Europe as well as the US have
techniques were used by Roy et al. to identify skin

14th ICCCNT IEEE Conference

Authorized licensed use limited to: Zhejiang University. Downloaded on June 30,2024 at 13:20:01 UTC from IEEE Xplore. Restrictions apply.
July 6-8, 2023
IIT - Delhi, Delhi, India
 IEEE - 56998

conditions such chicken pox, candidiasis, cellulitis, number of nations, with Nigeria witnessing the
and acne [3]. In 2020, a deep learning-based study greatest outbreak in 2017.
was performed using a tiny set of data consisting of
COVID-19 patients (108-infected and 86-not
infected), wherein 10 such distinct models were
examined by Ardakani et al, the accuracy of which
reached 99%, which was deemed commendable [4].
Recently, the year 2022 also witnessed a
convolutional neural network (CNN) based study,
suggested by Sanddep et al. They proposed that a
less complex CNN model can be used to diagnose
melanoma, psoriasis, chickenpox, and other skin-
related infections and that utilizing the currently
present VGGNet, around 71% of accuracy can be
achieved through picture analysis [4]. Their
suggested technique, in contrast, yields the best
outcomes by averaging an accuracy of about 78%.
The rest of this paper traces: An overview of the Fig. 2: Regional Distribution of Definite patients suffering
monkeypox outbreak in Section 2 with the medical from Monkeypox, as of July 2, 2022
diagnosis and a range of its symptoms discussed in (Reproduced from [6])
Section 3. Clinical management and treatment
methods employed till date are covered in Section 4. Since May 2022, there have been findings of fresh
Moving onto the paradigm of machine learning, the cases of monkeypox virus in nations not earlier
challenges of data collection are described in prone to the infection, with a simultaneous and
Section 5. Recent trends, past research, ML models consistent reporting from the list of endemic nations
in this domain are briefed in section 6 and 7. Section too. The CDC aka Center for Disease Control and
8 summarizes and concludes this paper. Prevention had reported as many as 5783 cases as
of July 1, 2022, which were documented in more
II.ORIGIN & OUTBREAK
than 53 countries. The spread of monkeypox
Monkeypox infection was first discovered in 1958 infections in regions worldwide is shown
while monkeys were under study in a laboratory at graphically in Figure 2. Currently, the western
the three State Serum Institutes in Denmark, hemisphere and parts of Europe are home to the
Copenhagen and South Africa [5]. It is likely that majority of monkeypox cases. Many cases of
monkeypox has been present throughout sub- monkeypox and groups of it have been documented
Saharan Africa since long before humans contracted often in both the virus-prone and non-prone
the infection from an infected creature. After countries spread across a wide variety of
smallpox was finally eradicated in 1970, it became geographical zones. World Health Organization
apparent that smallpox-like illnesses were still (WHO) conducted a study named "2022
occurring in rural areas. This resulted in the Monkeypox Outbreak: Global Trends.", according
identification of monkeypox as a separate disease. to which it has been witnessed that there was a
Monkeypox was revealed to be the source for a small (2.4%) rise in reported incidents from the
group of cases in the Midwest US in the summer of week dated 24 October 2022-20 October 2022 to
2003. In 2003, US witnessed the initial wave of the one dated 31 October 2022-6 November 2022
monkeypox, and it was traced back to the infection (week 43 to week 44). The bulk of cases (91%)
in prairie dogs which had been imported from sub- documented internationally, according to the report,
Saharan nations [6]. Since then, monkeypox has were in the Region of the Americas.
gained attention as a disease of worldwide public This paper presents a detailed analysis of the works
health concern. After 2003, there have been done in the field of Monkeypox detection using
numerous instances of monkeypox recorded in a Machine Learning, along with the spread of

14th ICCCNT IEEE Conference

Authorized licensed use limited to: Zhejiang University. Downloaded on June 30,2024 at 13:20:01 UTC from IEEE Xplore. Restrictions apply.
July 6-8, 2023
IIT - Delhi, Delhi, India
 IEEE - 56998

Monkeypox virus, its outbreak, diagnosis and appears or even up to three days later. The rash will
clinical treatment. The paper also focuses on the initially occur on the face before spreading fast in a
availability of datasets which remain absent from centrifugal pattern across the body. It is common to
most review publications as scant resources are find lesions in the oral cavity which make it harder
available for the topic. The paper also compares the for an infected person to eat and drink, reducing
work of different authors and the novel techniques their ability to get nourishment. The skin lesions
used by them. create significant skin disruption, which raises
worries about subsequent bacterial skin infection. In
DIAGNOSIS
III. about 19% of the infected population which were
Monkeypox does not have clinical manifestation in not vaccinated too, bacterial skin infections were
the incubation period, hence, not found to be observed as a ramification of monkeypox
contagious in this duration. The manifestation and contraction [8]. Second week of infection witnesses
symptoms of monkeypox begin to appear from the the appearance of symptoms associated with
onset of the prodromal period. This has to do with gastrointestinal organs such as diarrhoea, nausea and
how optional viremia spreads from the lymphoid vomiting. These can make the infected individual
organs, which are thought to be the starting point, up severely dehydrated, and the ulcers increase their
to the third organs, such as the eyes, lungs, digestive risk of dehydration. The ocular infection is the most
systems, and so on, along with the skin. According severe complication of monkeypox among the
to some reports, an infected person is more others shown in Fig. 3. Infected eyes can develop
contagious at the prodromal period. corneal scarring and lose their vision permanently.
There are a number of potential modes through
which the monkeypox virus can spread, but they are
all thought to involve direct contact with infected
people or animals. Direct contact or exposure to
infected animals can lead to animal to human
transmission. Secreted fluids such as respiratory
excretions, saliva, or discharge from lesions on skin
or those in mucosal parts of the body are the primary
causes of the spread. Exposure to contaminated
animal faeces in endemic regions of Africa with
limited infrastructure and supplies, and consumption
of their meat can be a significant risk factor, raising
their likelihood of being exposed to monkeypox.
Fig. 3: Classification of Monkeypox Complications and Transmission from homo sapiens sapiens-to-homo
Nonspecific Symptoms sapiens sapiens is rarer than from animal-to-homo
(Reproduced from [6]) sapiens sapiens. This mode of transmission is
Fig. 3 shows the general, non-specific monkeypox airborne due droplets secreted via respiration
symptoms that appear after a person is said been processes which results from either and extended in-
exposed to the infection for almost one to two person proximity or with the lesions of the
weeks. In instances before 2022, it had been contaminated individual. It has also been discovered
observed that the rashes took up to three days to that men with an history of sexual activity with other
show up distinctively after the onset of fever and men are more likely to catch the sickness during the
lymphoid growth. In the advent of the news of fresh current monkeypox outbreak. Monkeypox may be
infections, it was observed by professionals that for transferred by intimate physical contact, although
some infected group of people, the prodromal the WHO aka World Health Organization is not yet
symptoms may be minimal or not even noted. This certain if it is to be considered under the list of
led to the inference that until the development of a sexually transmitted diseases (STD), due to lack of
rash, a portion of population might remain unaware evidence.[3].
of any visible symptomology in the least. Typically,
the fever will start to go down the day after the rash

14th ICCCNT IEEE Conference

Authorized licensed use limited to: Zhejiang University. Downloaded on June 30,2024 at 13:20:01 UTC from IEEE Xplore. Restrictions apply.
July 6-8, 2023
IIT - Delhi, Delhi, India
 IEEE - 56998

IV. CLINICAL MANAGEMENT & TREATMENT simply available. It has been discovered when given
There have been some recommendations from the enough training data, deep-learning systems have
CDC aka Centers for Disease Control and the potential to automatically and successfully detect
Prevention stating that monkeypox virus has no lesion on skin. Because monkeypox is so
permanent treatment. The CDC approved two uncommon, there had already been gap in
medicinal drugs, ‘Tecovirimat’ and ‘Brincidofovir’ knowledge among medical experts in all the various
prescribed to be taken orally, which had proven nations, prior to the present outbreak. In their quest
effective to cure the patients of smallpox but are to solve this difficult problem, medical professional
now being utilised to treat the monkeypox virus in are inspired by the accomplishments of supervised
order to meet the urgent need. Vaccination is the machine learning in the discovery of COVID-19.
best defence against the monkeypox virus, though. The scarcity of monkeypox-affected photographs of
Vaccines have been created to tackle the skin makes it difficult to use machine learning to
monkeypox virus, such that they are readily identify the disease from patient skin scans.
available too. Even though they have received FDA DATA COLLECTION
V.
approval, United States has not notarized any use of
it until now. Referring to the statistics of smallpox It is essential to find patients in the period of the
and its treatments it was concluded that vaccines fast-spreading monkeypox illness in many countries
such as ‘vaccinia’, and drugs such as ‘tecovirimat’, with monkeypox symptoms. Many clinical experts
‘cidofovir’, could prove potential in the management believe that artificial intelligence (computer-based
of monkeypox virus [9]. In 2022, as per the WHO, intelligence) systems may handle visual information
The European Medicines Agency (EMA) granted a to lessen the importance of clinical findings with
green flag to the vaccine ‘tecovirimat’ was the events. Emergency clinics in China and Italy
originally created to treat smallpox, to be utilized in sent computer-based intelligence-based and picture-
defiance of monkeypox [3]. Since tecovirimat is not handling-based translators at the beginning of the
currently readily accessible, any usage of it needs to Coronavirus, which is what we saw, in order to
be closely supervised. A diversity of viruses can be improve emergency clinics' competency in caring
tackled using ‘cidofovir’ which implements the for Coronavirus patients [1]. However, at the time
mechanism blocking the DNA polymerase. of writing, there were no currently accessible
Additionally, there have been reports of numerous datasets on monkeypox, which prevents using
orthopoxviruses, including ectromelia, vaccinia, artificial intelligence to effectively treat, examine,
cowpox, variola, monkeypox and rabbitpox having and avoid the infection. As supported by many
explicit sensitivity to ‘tecovirimat’. Tecovirimat, has
referred to writing that recently thought of restricted
potential therapeutic benefits for treating
dataset in creating artificial intelligence-based
monkeypox. In addition to the previously mentioned
treatment options, several classes of therapy, like models during the beginning of Coronavirus
symptomatic and supportive, have been found to be illnesses, our underlying dataset contains
essential for containing the spread of the exceptionally restricted examples that won't be a
monkeypox infection [9]. It's critical to realise that problem for the underlying trial and error.
for monkeypox, there is no proven 100% effective
treatment, except from managing the symptomology
and avoiding complications. There is a need for
more research prior to any therapy is recommended
or vaccine can be developed in light of the current
outbreak of cases worldwide. Due to the virus's
resemblance to measles and chickenpox, it can be
difficult to diagnose monkeypox early in the course
of the illness. Computer-assisted abrasion structural
identification may help in the overseeing and speedy
diagnosis of monkeypox patients when confirmatory
polymerase chain reaction (PCR) procedures are not Fig. 4: Sample set of images from our referred data

14th ICCCNT IEEE Conference

Authorized licensed use limited to: Zhejiang University. Downloaded on June 30,2024 at 13:20:01 UTC from IEEE Xplore. Restrictions apply.
July 6-8, 2023
IIT - Delhi, Delhi, India
 IEEE - 56998

(Reproduced from [1]) classes of contamination, then differentiated their

findings to those of the VGG-16 set-up model. [11]
After exploring the available data sets, we found the Their approach gave the location a 78% accuracy.
dataset in [1] as the most appropriate dataset. The Additionally, Glock et al. performed measles illness
dataset is divided into many sub-organizers that identification using an exchange learning strategy
include both increased and decreased informative [12]. By applying the ResNet-50 model to the
indices. We utilize them in our evaluation since DL dataset of different rash images, they were able to
models prefer larger photos to learn key info even achieve a responsiveness of 81.7%, specificity of
more precisely. The dataset has the following 97.1%, and accuracy of 95.2%.
numbers mentioned in Table 1. Recently, Ahsan et al. gathered pictures of clerics
with various categories of diseases using web
Table I: Dataset in tabulated form mining technologies and had the images verified by
specialists. [1]. Later, they used the VGG-16 model
Disease name No. of images in data to evaluate an exchange learning technique while
Chickenpox Cases 329 considering two approaches. The first step
Measles Cases 286 investigated categorising the images into two groups
Monkeypox Cases 587 of illnesses: chickenpox and monkeypox. The
Normal Cases 552 second step extended the image set. Without
Total 1754 information expansion, they were able to group the
monkeypox with an accuracy of (97%) but with
expansion, the accuracy dropped to (78%). They
VI. RECENT TRENDS AND RESEARCH IN THE chose pre-trained Dl models like VGG, Inception
FIELD V3 and Xception, MobileNet and Efficient-Net. The
implementation was done using Keras in Python
Monkeypox virus came into picture in 1970.
[13, 14]. Once the boundaries have been adjusted,
However, the disease came into major existence as
for example, the picture size and increase as well as
an outbreak in recent times. A growing area of
rescale, pivot length, breadth change span, level
artificial intelligence (AI), known as machine
shift span, sheer range, zoom reach, and channel
learning (ML), has demonstrated exceptional
change scope [11]. The probabilistic values were
dedication in a variety of contexts, including
extracted by them from every calibrated model
dynamic assistance, contemporary applications,
(which were pre prepared/trained) and used the
clinical imaging, and disease detection. There are
larger part casting a ballot method to group the
very few research and reviews on the detection of
different DL models. They selected the 2 foremost
the disease using computer-based intelligence as a
fine-tuned models: Xception and DenseNet169 in
whole. We came across a few publications and
reference to the empirical study. After running all
studied and reviewed the models used by them in
the models, they found out that Xception came out
their detection models. A profound learning model
to be runner up among the DL models which were
(Res-COvNet) was promoted by Madhavan et al
pre-trained with an accuracy of 86.52%. Whereas,
[10] for the purpose of identifying coronavirus
the ensemble proposed by them proved to be the
infections. They expanded the structure with a
best performing with an accuracy of 87.13% [11].
classification layer after using ResNet50 to
For explaining their results, they used Grad-CAM
eliminate the elements from X-beam images. With a
(Gradient-weighted Class Activation Mapping) [15]
promising precision of 96.2%, their method
and LIME (Local Interpretable Model-Agnostic
successfully identified usual, respiratory disease,
Explanations) [16]. The LIME is a neighbourhood
viral tuberculosis, and Covid cases on X-bar images.
model-free method to deal with produce the
[11] Some other works revolved around using DL
understanding for a specific situation, transforming
models to identify diseases such as Chickenpox,
the data into a flow of comprehensible
Herpes, etc. Sandeep et al developed. They trained a
neighbourhood representations.. The Graduate CAM
CNN viz Convolutional Neural Network to
estimates the slope of the result include map of a
categorise the epidermis sore into eight (8) different
chosen layer of the organization. Comparing

14th ICCCNT IEEE Conference

Authorized licensed use limited to: Zhejiang University. Downloaded on June 30,2024 at 13:20:01 UTC from IEEE Xplore. Restrictions apply.
July 6-8, 2023
IIT - Delhi, Delhi, India
 IEEE - 56998

chickenpox and common images to measles and Similarly, diseases as FLD, aka, Fatty Liver Disease,
monkeypox illness, their method can distinguish can be detected by using Logistic Resression, ROC,
chickenpox and common images very well. Naïve Bayes classification etc. This is evident in
Additionally, their proposed model correctly [19]. So, many authors tried a similar thing for
categorises all occurrences of chickenpox - 66 in monkeypox as well. For monkeypox, the clinical
count and typical images - 111 in count in the test tests PCR aka Polymerize Chain Reaction tests are
set, but the segregation of measles and monkey-pox required for confirmation. But PCR tests are not
infections is still flawed [11]. readily accessible. Therefore, ML would aid in
detection of monkeypox. Hence, authors in [17],
developed two algorithms namely, binary PSOBER
algorithm, and SCBER algorithm. Basically, they
started with feature selection techniques and also
used data augmentation for dataset for validation as
well as training sets. Feature Engineering include
highlighting the key components that AI pipelines
will need. The primary aspect of component design
is included determination since it determines which
details will be streamlined for execution. Each
component of the collection of n-highlights To
indicate whether anything is crucial for the most
recent arrangement, the worth might be lowered to 1
or 0 in the element selection process [20]. Examples
of deep networks are VGG, AlexNet, GoogleNet
and ResNet. Two optimization algorithms were put
forward to improve feature selection. The primary
computation, referred to as PSOBER, is based on
Al-Biruni Earth span streamlining calculations and
Fig.5 Classification using GradCAM and LIME method molecular swarm advances. For feature selection,
(Reproduced from [11]) the twofold version of this computation is used. The
next calculation, referred to as SCBER, was applied
The paper concluded that the best-performing DL to Al-Biruni Earth radius (BER) enhancement
models were distinguished through examination calculation and the SCA calculation (Sine Cosine).
utilizing approved assessment measurements, and For a better choice of the optimal layout, their
they were then joined for worked on generally proposed calculations take advantage of the
execution and the ensemble model proved to be the advantages of the two computations they apply.
most accurate (Precision- 85.44%; Recall- 85.47%; Three experiment sets were performed by them on
F1-score- 85.40%; and Accuracy- 87.13%) [11]. their two proposed algorithms. Parameters such as
accuracy, sensitivity, specificity etc. were increased
In the year 2022, Al Biruni Earth Radius after using feature selection. Accuracy showed an
Optimization algorithm was used to classify increase of 3% after using feature selection method
MonkeyPox images. Authors in [17] review the along with the algorithms used by them [17].
models used in distinct diseases like Parkinson’s
disease, Alzheimer’s disease, some diseases related Apart from the approaches discussed above, Md
to chest, kidney etc. At this time, there is no Manjurul Ahsan, Muhammad Ramiz Uddin and
validated symptomatic test for Alzheimer's disease. their co-authors used a method called transfer
The affected person’s clinical past, the reports of learning. Transfer learning is referred to as an AI
mental and laboratory tests, and an methodology wherein we create models for one task
electroencephalogram (in some cases) should all be and it is used as the basis for a model for another. It
considered before making a decision. However, is a notable deep learning strategy where pre-
authors in [18] used SVMs i.e., Support Vector arranged models are utilized as the underlying stage
Machines to make the work of EEG epochs easier. on PC vision and ordinary language taking care of

14th ICCCNT IEEE Conference

Authorized licensed use limited to: Zhejiang University. Downloaded on June 30,2024 at 13:20:01 UTC from IEEE Xplore. Restrictions apply.
July 6-8, 2023
IIT - Delhi, Delhi, India
 IEEE - 56998

ventures because of the gigantic cycle and time

assets expected to cultivate cerebrum network
models on these issues and from the huge jumps in
mastery they give on related issues. Their model had
3 elements namely, pre preparedstructural design,
layer, and and an estimate group. They obtained the
prediction layer from [17].

Fig. 7: Pipeline implemented for skin lesion monkeypox

detection (Reproduced from [2])

Authors in [2] also proposed a full-fledged

application to detect monkeypox. The pipeline made
by them is shown in the figure. In their research,
they presented their core findings on how to identify
monkeypox skin sores using well-known deep-
learning architectures. However, their research had
some disadvantages as well. Their dataset has a cap
on the number of new patients who were added
Fig. 6: Monkeypox classification as proposed in [17]
(Reproduced from [17]) throughout time. As a result, their models' potential
for speculation is reduced. They utilized the loads
They also used LIME model aka Local interpretable that were pre-prepared from the ImageNet database
model agnostic explanations. By using a LIME for transfer learning as their underlying approach;
model to approximate any black box machine this dataset does not contain any images of skin
learning model, this technique can explain each sores. Thusly, utilizing a multi-source dataset of
individual prediction. In their work, they firstly dermatoscopic picture information to pre-set up the
collected and made a publicly monkeypox dataset models could additionally improve the structure's
and also made a modified version of the VGG16 exactness and generalizability.
model. They put out and evaluated a modified VII. MODELS WHICH CAN BE USED FOR
VGG16 model that inculcates the results of two (2) DETECTION OF MONKEYPOX
different studies, aka Study One (1) and Study Two
(2). Their preliminary computational findings A. DL Models
demonstrated that the model suggested by them can The accessibility of several DL models built upon
distinguish people suffering from Monkeypox with the enormous ImageNet dataset has incredibly
an exactness equal to 97.2% (AUC equals 97.2) and advanced picture categorization and PC vision tasks.
88.8% (AUC equals 0.867), respectively, for Studies More clearly still, using these pre-built DL models
One and Two. [2] for move learning is a common methodology when
access to master marked information is restricted to
specific areas, such as biomedical picture analysis.
Since move learning enables DL models built upon
massive datasets to move learned information to a
small-scale space explicit dataset, this is helpful to

14th ICCCNT IEEE Conference

Authorized licensed use limited to: Zhejiang University. Downloaded on June 30,2024 at 13:20:01 UTC from IEEE Xplore. Restrictions apply.
July 6-8, 2023
IIT - Delhi, Delhi, India
 IEEE - 56998

support the presentation in a scenario with limited algorithm. The PSO strategy relies on imitating the
information. There are many pre-trained models, we scavenging behaviour of rushing species, like birds,
will be discussing the two best performing ones using an augmentation tool that outsmarts the
below. method of obtaining knowledge. The BER
computation functions in even-number cycles
during the search cycle. The PSO computation
operates in cycles of odd numbers. The improved
investigation and double-dealing of the search cycle
are taken into consideration in this trade activity of
the BER and PSO. Basically, apart from PSOBER
model, a binary PSOBER model was also proposed.
The two double qualities 0 and 1 make up the whole
small inquiry space that feature selection problems
use to determine if a particular feature is
noteworthy [11]. The twofold PSOBER approach is
next put forward in this section in order to adapt to
the feature selection methodology by changing the
Fig.8: Process to train the pre-trained DL models constant qualities produced by the persistent
(Reproduced from [11]) PSOBER computation into double [0, 1].
1) VGG-16: C. SCBER Model:
VGG is for Visual Math Gathering, and it is a The then definedSCBER model is an amalgamation
typical, multi-layered, profound Convolutional of the Sine Cosine algorithm - SC and the Al Biruni
Neural Network (CNN) design. It consists of 16
or 19 convolutional layers (VGG-19) [21]. The Earth Radius Algorithm - BER. Since the SCA
foundation of pivotal article acknowledgment computation uses just one ideal solution for many
models is the VGG design. developed as a diverse layouts, it is more robust than a wide radius
sophisticated brain arrangement, The VGGNet of meta-experimental works in the past. This
also performs better than baselines on several calculation's combined speed and memory use are
tasks and datasets other than ImageNet. extremely low when compared to other calculations.
Additionally, it remains one of the most used The borders of a multi-facet brain structure are
photo acknowledgment formats at the moment.
streamlined using the SCBER enhancement
computation [11]. When used on a populace, the
2) Xception:
suggested SCBER computation discovers some
Xception is basically a convolutional neural
degree of harmony between looking into and
network aka CNN which is 71 layers deep. A
profiting from different segments of the population.
pretrained version of the network built using more
than 1,000,000 images from the ImageNet data set VIII. CONCLUSION
can be stacked. The prepared network can
Monkeypox, being a viral infection, has no specific
characterize pictures into 1000 unique item
cure to it. The viral disease may prove lethal in
classifications, including console, mouse, pencil,
some cases. They may be with or without co-
and different animals [22-28]. The network has
morbidities. We firstly assessed and reviewed the
since learned rich component portrayals for a lot of
research made in the field of virology for diagnosis
images.
of the virus. We also went through and reviewed the
B. PSOBER Model: available data-sets that can be used for monkeypox
The PSO algorithm and the BER algorithm prediction using machine learning. Not many data-
combines to form this hybrid PSOBER Model. PSO sets are available for this purpose, so the
is the Particle Swarm Optimization algorithm. BER implemented models yet are not that efficient due to
is the Al-Biruni Earth radius optimization lack of training. In this paper, we discussed and

14th ICCCNT IEEE Conference

Authorized licensed use limited to: Zhejiang University. Downloaded on June 30,2024 at 13:20:01 UTC from IEEE Xplore. Restrictions apply.
July 6-8, 2023
IIT - Delhi, Delhi, India
 IEEE - 56998

reviewed some of the many techniques which can [6] Kaler J, Hussain A, Flores G, et al. (July 03, 2022)
be used for easier and earlier detection of Monkeypox: A Comprehensive Review of Transmission,
Pathogenesis, and Manifestation. Cureus 14(7): e26531.
monkeypox, so that its infection can be stopped doi:10.7759/cureus.26531
from spreading. We discuss VGG16 model, [7] 2022 global map & case count.
PSOBER model, SCBER model, Xception model https://www.cdc.gov/poxvirus/monkeypox/response/
etc. The latter two models are the combinations of 2022/world-map.html.
few already existing models and are considered to [8] McCollum AM, Damon IK: Human monkeypox. Clin
Infect Dis. 2014, 58:260-7. 10.1093/cid/cit703
be superior due to the promising results they [9] Okyay, Ramazan & Bayrak, Edip & Kaya, Erhan & Sahin,
presented. We also reviewed the application Ahmet & Koçyiğit, Burhan & Tasdogan, Ali & Avci,
implemented to detect and determine if the skin Akkan & Sumbul, Hilmi. (2022). Another Epidemic in the
lesions are caused by monkeypox. Shadow of Covid 19 Pandemic: A Review of Monkeypox.
However, implementation work needs to be carried Eurasian Journal of Medicine and Oncology. 6. 95-99.
10.14744/ejmo.2022.2022.
out for improvement in accuracy, precision, F1 [10] Mangena, Venu & Khamparia, Aditya & Gupta, Deepak
Score and Recall of the present work done by the & Pande, Sagar & Tiwari, Prayag & Hossain, M.
other authors. More precisely data set is also a Shamim. (2021). Res-CovNet: an internet of medical
requirement. Some other models for feature health things driven COVID-19 framework using transfer
engineering purposes can be used in order to learning. Neural Computing and Applications.
[11] Sitaula, Chiranjibi & Basnet, Anish & Mainali, Ashish &
increase the efficiency of the current implemented Shahi, Tej. (2021). Deep Learning-Based Methods for
models such as in [11] and [12]. There is a future Sentiment Analysis on Nepali COVID-19-Related
scope of improvement in the accuracy of the deep Tweets. Computational Intelligence and Neuroscience.
learning models which will be carried out further by 2021. 10.1155/2021/2158184.
using techniques such as transformer and ensemble [12] Glock, Kimberly & Napier, Charlie & Gary, Todd &
Gupta, Vibhuti & Gigante, Joseph & Schaffner, William
approaches. & Wang, Qingguo. (2021). Measles Rash Identification
Using Transfer Learning and Deep Convolutional Neural
REFERENCES
Networks. 3905-3910.
[1] Ahsan, Md Manjurul & Uddin, Muhammad Ramiz & 10.1109/BigData52589.2021.9671333.
Farjana, Mithila & Sakib, Ahmed & Momin, Khondhaker [13] Chollet F, et al. (2015) Keras.
& Luna, Shahana. (2022). Image Data collection and https://github.com/fchollet/keras
implementation of deep learning-based model in detecting [14] Tan, Mingxing & Le, Quoc. (2019). EfficientNet:
Monkeypox disease using modified VGG16. Rethinking Model Scaling for Convolutional Neural
10.48550/arXiv.2206.01862. Networks.
[2] Ali, Shams Nafisa & Ahmed, Tazuddin & Paul, Joydip & [15] R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D.
Jahan, Tasnim & Sani, s.M. & Noor, Nawsabah & Hasan, Parikh and D. Batra, "Grad-CAM: Visual Explanations
Taufiq. (2022). Monkeypox Skin Lesion Detection Using from Deep Networks via Gradient-Based Localization,"
Deep Learning Models: A Feasibility Study. 2017 IEEE International Conference on Computer Vision
10.48550/arXiv.2207.03342. (ICCV), 2017, pp. 618-626, doi: 10.1109/ICCV.2017.74.
[3] Ahsan, Md Manjurul & Uddin, Muhammad Ramiz & [16] Ribeiro MT, Singh S, Guestrin C (2016) Why should I
Farjana, Mithila & Sakib, Ahmed & Momin, Khondhaker trust you?: Explaining the predictions of any classifer. In:
& Luna, Shahana. (2022). Image Data collection and Proceedings of 78 Page 8 of 9 Journal of Medical
implementation of deep learning-based model in detecting Systems (2022) 46:78 1 3the 22nd ACM SIGKDD
Monkeypox disease using modified VGG16. International Conference on Knowledge Discovery and
10.48550/arXiv.2206.01862. Data Mining, San Francisco, CA, USA, August 13-17,
[4] Ali Abbasian Ardakani, Alireza Rajabzadeh Kanafi, U. 2016, pp 1135–1144
Rajendra Acharya, Nazanin Khadem, Afshin Mohammadi, [17] Abdelhamid, Abdelaziz & El-kenawy, El-Sayed &
Application of deep learning technique to manage Khodadadi, Nima & Mirjalili, Seyedali & Khafaga, Doaa
COVID-19 in routine clinical practice using CT images: & Alharbi, Amal & Ibrahim, Abdelhameed & Eid,
Results of 10 convolutional neural networks,Computers in Marwa & Saber, Mohamed. (2022). Classification of
Biology and Medicine,Volume 121,2020,103795,ISSN Monkeypox Images Based on Transfer Learning and the
0010-4825, Al-Biruni Earth Radius Optimization Algorithm. 10.
https://doi.org/10.1016/j.compbiomed.2020.103795. 3614. 10.3390/math10193614.
[5] Marlyn Moore and Farah Zahra. Monkeypox [18] Mohan, Senthilkumar & Thirumalai, Chandra Segar &
https://www.ncbi.nlm.nih.gov/books/NBK574519/, May Srivastava, Gautam. (2019). Effective Heart Disease
2022.

14th ICCCNT IEEE Conference

Authorized licensed use limited to: Zhejiang University. Downloaded on June 30,2024 at 13:20:01 UTC from IEEE Xplore. Restrictions apply.
July 6-8, 2023
IIT - Delhi, Delhi, India
 IEEE - 56998

Prediction Using Hybrid Machine Learning Techniques. in Networks and Systems, vol 145. Springer,
IEEE Access. PP. 1-1. 10.1109/ACCESS.2019.2923707. Singapore. https://doi.org/10.1007/978-981-15-7345-
[19] J. Qin, L. Chen, Y. Liu, C. Liu, C. Feng and B. Chen, "A 3_47
Machine Learning Methodology for Diagnosing Chronic [25] Malti Bansal, Apoorva Goyal, Apoorva Choudhary,
Kidney Disease," in IEEE Access, vol. 8, pp. 20991- “A comparative analysis of K-Nearest Neighbor,
21002, 2020, doi: 10.1109/ACCESS.2019.2963053. Genetic, Support Vector Machine, Decision Tree, and
[20] El-kenawy, El-Sayed & Mirjalili, Seyedali & Alassery, Long Short Term Memory algorithms in machine
Fawaz & Zhang, Yu-Dong & Eid, Marwa & El-Mashad, learning”, Decision Analytics Journal, Volume 3,
2022, 100071, ISSN 2772-6622,
Shady & Aloyaydi, Bandar & Ibrahim, Abdelhameed &
https://doi.org/10.1016/j.dajour.2022.100071.
Abdelhamid, Abdelaziz. (2022). Novel Meta-Heuristic
[26] M. Bansal, T. Chopra and S. Biswas, "Organ
Algorithm for Feature Selection, Unconstrained Simulation and Healthcare Services: An Application of
Functions and Engineering Problems. IEEE Access. 1-1. IoT," 2021 6th International Conference on Inventive
10.1109/ACCESS.2022.3166901. Computation Technologies (ICICT), 2021, pp. 205-208,
[21] Deng, Jia & Dong, Wei & Socher, Richard & Li, Li-Jia doi: 10.1109/ICICT50816.2021.9358677.
& Li, Kai & Li, Fei-Fei. (2009). ImageNet: a Large-Scale [27] Malti Bansal and Varun Sirpal, “Fog Computing-based
Hierarchical Image Database. IEEE Conference on Internet of Things and its Applications in Healthcare”,
Computer Vision and Pattern Recognition. 248-255. 2021, J. Phys. Conf. Ser. 1916 012041
[22] Chollet, Francois. (2017). Xception: Deep Learning with [28] Bansal, M., Priya (2022). Machine Learning Perspective
Depthwise Separable Convolutions. 1800-1807. in VLSI Computer-Aided Design at Different Abstraction
10.1109/CVPR.2017.195. Levels. In: Shakya, S., Bestak, R., Palanisamy, R., Kamel,
[23] M. Bansal and Priya, "Application Layer Protocols for K.A. (eds) Mobile Computing and Sustainable
Internet of Healthcare Things (IoHT)," 2020 Fourth Informatics. Lecture Notes on Data Engineering and
International Conference on Inventive Systems and Communications Technologies, vol 68. Springer,
Control (ICISC), 2020, pp. 369-376, Singapore. https://doi.org/10.1007/978-981-16-1866- 6_6
doi: 10.1109/ICISC47916.2020.9171092. [29] https://www.nebraskamed.com/infectious-
[24] Bansal, M., Priya (2021). Performance Comparison of diseases/monkeypox/how-does-monkeypox-start-plus-5-
MQTT and CoAP Protocols in Different Simulation pictures-to-show-how-the-bumps-progress
Environments. In: Ranganathan, G., Chen, J., Rocha,
Á. (eds) Inventive Communication and
Computational Technologies. Lecture Notes

14th ICCCNT IEEE Conference

Authorized licensed use limited to: Zhejiang University. Downloaded on June 30,2024 at 13:20:01 UTC from IEEE Xplore. Restrictions apply.
July 6-8, 2023
IIT - Delhi, Delhi, India