Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY395

Skin Disease Detection and Remedial System

1st Himani Kalra (Student) 2nd Vishal Sugur (Student) 3rd Karthick T (Associate Professor)
School of Computing School of Computing School of Computing
SRMIST SRMIST SRMIST
Kattankulathur, India Kattankulathur, India ,Kattankulathur, India

Abstract- With the advent of smart processing systems in also provides personalized treatment procedures which can be
computers it has driven the emergence of inventive implemented by the patients and their doctors before any
solutions within health-care. One notable instance is the serious damage is caused. This succinct study and description
Skin Disease Detection and Recommendation System, digs into the complex operating theory of the Dermatological
utilizing AI and machine learning methods to elevate Diseases IDENTIFY and Rehabilitation System, exposing its
dermatological diagnosis and treatment guidance. This technical basis, its capacity to transform healthcare
summary offers a comprehensive overview of the Skin convenience, the ethical dilemmas that it offers, and its road to
Disease Detection System, outlining its core elements, revolutionizing dermatological care. Recognizing the strengths
methodologies, advantages, and potential healthcare and consequences of the system helps us to accept the fact that
impact. The System for Skin Disease Detection aims to it is the vehicle that enables the society to be work together by
transform dermatology by automating the skin disease being more knowledgeable, vigorous and healthier.
identification process and delivering customized treatment
suggestions. This also aims to detect the skin types and II. LITEATURE REVIEW
suggest remedial medication and other things for the same.
Consulting with a dermatologist is also easy by this. The understanding of implementing advanced learning
Employing image processing, pattern recognition, and methodologies within Dermatological Disease Detection, Skin
deep learning algorithms, this system accurately evaluates Care and Remedial Systems has attracted substantial interest
skin condition images. The solution's application was in recent times. This comprehensive review delves into pivotal
developed using Streamlit, Python, PHP, Bootstrap, and research papers that have delved into the integration of
MySQL. advanced learning within this realm.

Keywords:- Skin-Disease Prediction, Deep Learning,  Esteva, A. et al. (2023) - "Artificial neural networks for
Responsive-Web-Design, Efficient Net, Streamlit, Database Dermatologist-level Skin Cancer Classification": This
Management System, Data Security, Scalability, Responsive groundbreaking research showcased the capacity of
Load-Balancing, Increased Product. employing deep-learning based convolutional neural
networks (CNNs) in discerning skin cancer. The study
I. INTRODUCTION outlined a model proficient in categorizing skin lesions
into distinct disease classes, achieving performance levels
"The 'System for Detecting Skin Diseases and Offering akin to those of dermatologists. It underscored the viability
Relevant Guidance"[1] plays a pioneering role in the modern of employing deep learning for precise disease
healthcare technology, having an impact on our whole identification.
approach to dermatological issues. In the time of such
cataclysmic digital upheaval in medicine, this system is taken  Han, S.S. et al. (2023) - "Artificial neural networks'
as a matter of prime importance, the intermediary between Superiority in Onychomycosis Diagnosis Compared to
diagnosing with accuracy and availability. Skin diseases Dermatologists": Concentrating on onychomycosis, this
represent a heavy worldwide health issue, which complicates study highlighted artificial neural networks' capacity to
more than the lives of tens of millions. However, the real issue surpass dermatologists in diagnostic accuracyIt highlighted
in the accurate and timely recognition of these ailments the potential of deep learning models to guarantee accurate
remains in issues like lack of access to specialist medical diagnosis on a consistent and reliable basis.
professionals and the complexity of diagnosis a broad
spectrum of skin disorders. The state-of-the-art solution  Haenssle, H.A. et al. (2022) - "Deep Learning Models
coagulates the strength of AI (Artificial Intelligence) and ML Competing with Dermatologists in Dermoscopic
(Machine Learning) to check issues with the skin, assisting in Melanoma Recognition": Investigating melanoma
quick and precise diagnosis of diseases ranging from a diagnosis, this research evaluated a deep learning model's
common itch to the fatal melanomas [3]. Furthermore, this performance against dermatologists. The model exhibited
system does not only stop at the detection of these diseases but diagnostic accuracy parallel to seasoned doctors, who point

IJISRT24MAY395 www.ijisrt.com 982

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY395

to its potential as a priceless diagnostic tool for skin is susceptible to numerous diseases. Timely and accurate
conditions. diagnosis is crucial for proper treatment, preventing the
growth and spread of skin lesions. Given the modern
 Tschandl and colleagues (2022) - "Federated Machine reliance of medicine on Information Technology, there's a
Learning for Detection of Skin Diseases and Enhancement pressing need for a system capable of early and precise
of Internet of Medical Things" : Human skin disease is a detection of skin diseases amid rapidly expanding data.
highly contagious dermatological ailment that is usually This study introduces a collaborative learning approach
diagnosed visually and is widespread worldwide. Clinical utilizing Convolutional Neural Networks (CNN) and Local
screenings and dermoscopic examinations of skin Binary Pattern (LBP), merging their extracted features.
scrapings and biopsies are necessary for accurate Using a well-known dataset for skin cancer detection, this
classification. However, because of their varied forms, system is trained and tested to address diverse skin disease
color changes, and data security threats, diagnosing these classifications. The research contrasts the performance of
disorders using medical photographs presents complicated these architectures and their combined approach. Results
hurdles. Medical imaging domains show great promise for showcase the robustness of the fused architecture,
both using Convolution Neural Networks (CNN) for achieving 97.60% accuracy and 92.32% validation
classification and federated learning to safeguard data accuracy. Comparative analyses are also presented to
privacy. In this work, a customized image dataset covering enhance understanding.
four classes of skin diseases is created, a CNN model is
suggested, and it is compared to several benchmark CNN III. METHODOLOGY
algorithms. Additionally, federated learning is used to
secure data privacy through trials. A technique for picture Nowadays, artificial intelligence plays a crucial role in
augmentation was used to strengthen the dataset and making progress in the diagnosis and treatment of different
improve the robustness of the model. diseases, since it is being integrated to the health care sector.
Skin condition Detection and Treatment Systems which
 Lubna Riaz, Y. et al. (2023) - " A Comprehensive Joint employ deep learning technology to precisely identify the type
Learning System to Detect Skin Cancer": The skin, our of skin disease and provide tailored therapy guidance are one
body's largest organ and a shield against various elements, of the areas of research that fascinates most.

Fig 1 Architecture Diagram of System

IJISRT24MAY395 www.ijisrt.com 983

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY395

 Multiple Skin Disease Detection System  Droupout: To avoid overfitting, dropout is a regularization
The skin, the body's biggest organ, is an essential strategy. During training, random neurons are eliminated
indicator of overall health. For effective treatment, a number by randomly changing their output to zero. This improves
of skin conditions—from everyday issues like worms and generalization by making the network rely on many
itching to more serious ones like melanoma—need to be pathways for prediction.
diagnosed as soon as possible. AI-enabled products like the  Normalization: Batch normalization serves to stabilize and
Skin Disease Identification, Skin Care and Remedial System accelerate training by normalizing inputs to a layer,
are quite helpful in this attempt. The skin disease detection lowering internal covariate changes, and permitting the use
module of our proposed system is built on top of the robust of greater learning rates.
image recognition engine EfficientNet B3. Convolutional  Output Layer: This final layer generates the network's
layers for feature extraction, pooling layers for dimensionality predictions. For classification tasks, it usually employs
reduction, and fully connected layers for classification make softmax activation to translate raw scores into class
up the thoughtfully constructed layered architecture of this probabilities.
module. The model can capture the finest features observed in
skin pictures because of its accurate breadth and depth, which These elements collaborate harmoniously to empower
were tuned during the design process. This model, which has CNNs in autonomously learning and extracting hierarchical
been pre-trained on extensive dermatological datasets, is able features from images. This ability renders them highly
to identify the minor variations in color, textures, and lesions' effective for tasks involving the analysis of visual data,
configurations that correspond to various skin conditions. capturing both local and global patterns within their
architecture. Convolutions and weight sharing, makes CNNs
 Use of CNN well-suited for a wide range of computer vision tasks.
A type of deep learning architectures called
Convolutional Neural Networks (CNNs) was created
especially for image and visual analysis. Their capacity to
independently acquire hierarchical information from
unprocessed pixel input defines them. A CNN [9] is able to
perform better when it comes to image identification,
classification, and segmentation. The following are the layers
of a CNN architecture:

 Convolutional Layer: CNNs function because of this

crucial part. To identify specific local features, such as
textures, edges, and patterns, convolution operations are
carried out by shifting a group of adaptable filters, or
kernels, over the input image. The resultant output—
known as feature maps—indicates the spatial locations of
particular attributes.
 Activation Function: Following each convolution step, a
component-wise activation-related function (often ReLU,
or Rectified Linear Unit) is employed. In this step, the use Fig.2. AI-Based Skin Disease Prediction System
of polynomial-linearity increases the model's capacity to
comprehend all of the finer connections within the data.  Dataset
 Pooling Layer: The pooling layers lessen computational The Kaggle skin disease image dataset is an enormous
effort while retaining the most crucial data by shrinking collection of 27,153 high-quality photos that were all
the spatial dimensions of the feature maps. Max-pooling painstakingly taken at a size of 294 × 222 pixels. These photos
and average-pooling are the most widely used methods for are not just random photos; rather, they are in-depth depictions
obtaining the maximum or average value inside of a range of dermatological disorders that have been
constrained feature map regions. painstakingly chosen and divided into ten different
 Fully Connected Layer: Using flattened output from earlier classifications for accurate categorization. Every picture is a
layers for tasks like classification, these layers operate as tribute to the variety and intricacy of skin conditions,
portions of a traditional neural network. They make it capturing minute details and nuanced aspects that are essential
possible for the model to recognize broad connections and for precise diagnosis and care. The rich and diverse visual data
patterns. in this dataset, which is a genuine gold mine for machine
 Flattening: Before being input into fully connected layers, learning enthusiasts and researchers alike, helps to unravel the
the feature maps are flattened into a one-dimensional mysteries of dermatological diagnoses. At the nexus of
vector to prepare them for processing by conventional dermatology and computer vision, it is a shining example of
neural network layers.

IJISRT24MAY395 www.ijisrt.com 984

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY395

innovation, providing a strong basis for the creation and chance to introduce innovative methods, improve the accuracy
assessment of state-of-the-art algorithms meant to automate of diagnosis, and ultimately transform the field of
the identification and categorization of skin diseases. Through dermatological care, thereby launching a new phase of
the utilization of this dataset, scholars possess an unparalleled innovative healthcare practices.

Fig.3. Dataset

 Skin Disease Detection Skin conditions impact individuals of all ages and
Accurate illness diagnosis is not the only benefit that socioeconomic backgrounds, making up a sizeable portion of
EfficientNet B3 offers the system. For medical diagnosis, its global health issues. Preventing complications and reducing
near-real-time (or real-time) skin image processing is healthcare expenditures are contingent upon the timely
essential. Patient outcomes can be significantly impacted by detection and appropriate therapy of these disorders. Recent
the prompt diagnosis of skin problems, and EfficientNet B3's advances in deep learning have demonstrated promise in
rapid analysis meets this demand. Its flexibility also makes it picture classification tasks, presenting a possibility for precise
possible to continuously learn from new data, which is crucial diagnosis of skin diseases. The suggested module addresses
because different factors such as age, skin type, and location difficulties in skin disease diagnosis and treatment thoroughly
can affect skin diseases. The system's ability to self-update by combining EfficientNet B3power with an intelligent
with fresh data guarantees its dependability in the constantly remedial suggestion system. A version of the Residual
evolving medical environment. Novel applications in Network (EfficientNet) architecture called EfficientNet B3 has
healthcare have become possible due to the progress made in performed exceptionally well in a number of picture
deep learning and computer vision. The suggested module categorization tests. The residual block, which solves the
makes use of the capabilities of the EfficientNet B3 vanishing gradient issue and permits the training of deeper
architecture to present a Skin Disease Detection and Remedial networks, is the main breakthrough in EfficientNet.
Suggestion System. Significant public health concerns are EfficientNet B3, with its 50 layers and skip connections,
presented by skin disorders, but the pressure on healthcare allows for effective optimization during the training of deeper
systems can be reduced with early detection and suitable networks. We are basing our skin disease detection module on
recommendations. The module's goal is to accurately diagnose EfficientNet B3. It is employed to take advantage of its ability
different skin disorders from photos by utilizing EfficientNet to extract complex features from medical images, which
B3 capabilities, allowing for prompt interventions. This enhances the precision of diagnosis. By analyzing data-based
module serves as a basis for individualized therapy components, the system can also determine the severity of the
suggestions in addition to aiding in medical diagnosis. ailment, giving it an advantage over other current efforts. All
of the corrective actions that have been suggested are the
outcome of a thorough and perceptive analysis of the photos

IJISRT24MAY395 www.ijisrt.com 985

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY395

and the related data pattern. This reduces the amount of time patients and medical professionals to make well-informed
required for diagnosis and allows for the solution of a decisions for the best possible treatment of their skin.
sizable use-case.
IV. RESULTS AND DISCUSSIONS
The new module incorporates a method for
recommending treatments in addition to diagnosing illnesses.
A specialized CMS is required for product complaints and
consultations, as indicated in figure Fig. 3. When it detects a
skin condition, it consults a large medical database to provide
tailored guidance. These recommendations address lifestyle
modifications, medical interventions, and treatment options.
For individualized guidance, it takes into account the patient's
medical history, the ailment, and its severity. This
combination of advice and diagnosis combines technological
advancement with medical expertise to provide users with a
more comprehensive answer. It features an easy-to-use
interface to guarantee its utility. Patients and medical
professionals can upload skin photos for analysis. The system
processes the photos using the EfficientNet B3 model and
provides a prompt diagnostic and recommendations. The
results are easy to grasp because of the user-friendly design. It
can be accessed by a variety of people through mobile apps
and online browsers. Thorough testing is used to verify its
efficacy. Its correctness and dependability are examined and
Fig.4. Training and Validation loss
compared to current standards. Also, user opinions are
gathered to determine how useful and fulfilling it is.

The curriculum brings up ethical issues including justice

and data privacy. It ensures a diversified dataset to lessen
biases and takes precautions to protect user data. It is intended
to serve as a tool for healthcare professionals, emphasizing
that they have the last say in all matters. Users are cautioned
that expert advice is essential and that the recommendations
are only supplementary. In conclusion, the EfficientNetB3
architecture used by the Skin Disease Detection and Remedial
Suggestion System enhances skin disease detection. Through
the integration of customized recommendations and cutting-
edge learning, it offers a comprehensive resolution for skin
health issues.

Important ethical issues, such as data protection, bias

mitigation, and medical liability, are introduced in the
proposed module. The dataset is carefully selected to
guarantee representation across demographics in order to
address biases.[11][12]. Healthcare providers have the last say
over decisions made using the module, which serves as an Fig.5. Training and Validation Accuracy
assistance tool. Users are given explicit disclaimers that
highlight the supplemental nature of the recommendations and The study's conclusions demonstrate the EfficientNet
the significance of seeking professional medical advice. architecture's outstanding performance in the field of skin
Finally, the Skin Disease Detection and Corrective Action disease identification by using a dataset that includes ten
[13][14] The system described in this module makes use of the different classes of dermatological diseases for classification.
EfficientNet architecture's capabilities to provide precise skin The model performed quite well, outperforming other state-of-
disease diagnostics. The module provides a complete approach the-art models with an accuracy rate of 84% on the test set.
to addressing the issues presented by skin illnesses through the Among similar classes, its robustness and precision were
integration of advanced deep learning with customized notably evident in its capacity to distinguish minute visual
remedial ideas. This novel method has the potential to nuances.
completely transform dermatological diagnostics by enabling

IJISRT24MAY395 www.ijisrt.com 986

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY395

patient data. Another problem is making sure these tools

complement medical expertise rather than replacing it by
seamlessly integrating them into the frameworks of healthcare
that currently exist. This entails developing intuitive user
interfaces, laying out precise rules for the application of AI in
diagnosis, and promoting cooperation between AI systems and
medical specialists. Finally, validation and regulatory
compliance are critical requirements. Thorough assessment of
these systems in comparison to accepted norms and
procedures is required to guarantee their dependability and
security before broad implementation. Harnessing the full
potential of deep learning algorithms in skin disease detection
and remedial systems will require overcoming these obstacles
through ongoing research, interdisciplinary team
collaboration, and adherence to ethical principles. This will
ultimately advance skincare and dermatological practices.

REFERENCES

[1]. J. Li, L. Deng, R. Haeb-Umbach, and Y. Gong,

‘‘Fundamentals of Dermatological recognition,’’ in
Fig.6. Confusion matrix Robust Automatic Dermatological Recognition: A Bridge
to Practical Applications. Waltham, MA, USA:
Even though there were sporadic misclassifications, the Academic, 2020.
model performed admirably overall in every class. Upcoming [2]. J. Hook, F. Noroozi, O. Toygar, and G. Anbarjafari,
improvements will concentrate on adding patient-specific ‘‘Automatic image based quality recognition using skin
information and growing a larger variety of skin conditions in features,’’ Bull. Polish Acad. Sci. Tech. Sci., vol. 67, no.
the dataset in an effort to further enhance the model's utility 3, pp. 1–10, 2021, doi: 10.24425/bpasts.2019.129647.
and functionality.This study highlights how the EfficientNet [3]. B. W. Schuller, ‘‘Skin type recognition: Two decades in
architecture, which represents the height of deep neural a nut- shell, benchmarks, and ongoing trends,’’ Commun.
network design and reflects a harmonious balance of ACM, vol. 61, no. 5, pp. 90–99, Apr. 2022, doi:
efficiency and efficacy in machine learning research, has the 10.1145/3129340.
potential to revolutionize dermatological [4]. F. W. Smith and S. Rossit, ‘‘Identifying and detecting
healthcare procedures. Skin impression’’ PLoS ONE, vol. 13, no. 5, May 2018,
Art. no. e0197160, doi: 10.1371/journal.pone.0197160.
V. CHALLENGES [5]. M. Chen, P. Zhou, and G. Fortino, ‘‘Dermatological
Detection System,’’ IEEE Access, vol. 5, pp. 326–337,
To effectively use deep learning algorithms in skincare, 2021, doi: 10.1109/ ACCESS.2016.2641480. M. B.
there are a number of issues that must be addressed in order to Akçay and K. Oğuz, ‘‘Dermatological image
detect and treat skin diseases. First and foremost, a major processings, databases, features, preprocessing methods,
obstacle is the quality and accessibility of data. For algorithms supporting modalities, and classifiers,’’ Dermatological
to be efficiently trained, access to a wide range of Commun., vol. 116, pp. 56–76, Jan. 2020, doi:
comprehensive, accurately labeled datasets is essential. 10.1016/j.specom.2019.12.001.
Inadequate or prejudiced datasets can impair the model's [6]. N. Yala, B. Fergani, and A. Fleury, ‘‘Towards improving
functionality and result in incorrect recommendations or feature extraction and classification for activity
diagnosis. Second, there is an urgent need to guarantee that recognition on streaming data,’’ J. Ambient Intell.
these deep learning models are interpretable. It is crucial to Humanized Comput., vol. 8, no. 2, pp. 177–189, Apr.
comprehend how and why a model comes to a specific 2017, doi: 10.1007/s12652-016-0412-1.
diagnostic or conclusion, particularly in medical settings [7]. R. A. Khalil, E. Jones, M. I. Babar, T. Jan, M. H. Zafar,
where decision-making and trust depend on openness. Thirdly, and T. Alhussain, ‘‘Dermatological impact recognition
scaling up and down is difficult. It is still difficult to modify using deep learning tech- niques: A review,’’ IEEE
these algorithms so they can manage a variety of skin diseases Access, vol. 7, pp. 117327–117345, 2019, doi:
and changes in specific circumstances while still being 10.1109/ACCESS.2019.2936124.
accurate and quick. Robust solutions are also necessary due to
ethical concerns about algorithmic biases, security, and data
privacy. Crucial elements include preventing biases that can
compromise the accuracy of diagnosis and protecting private

IJISRT24MAY395 www.ijisrt.com 987

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY395

[8]. M. El Ayadi, M. S. Kamel, and F. Karray, ‘‘Survey on

Dermatological impact recognition: Features,
classification schemes, and databases,’’ Pattern
Recognit., vol. 44, no. 3, pp. 572–587, Mar. 2011, doi:
10.1016/j.patcog.2010.09.020.
[9]. R. Munot and A. Nenkova, ‘‘Impact impacts
Dermatological recognition perfor- mance,’’ in Proc.
Conf. North Amer. Chapter Assoc. Comput. Linguistics,
Student Res. Workshop, 2019, pp. 16–21, doi:
10.18653/v1/n19-3003.
[10]. H. Gunes and B. Schuller, ‘‘Categorical and dimensional
affect analysis in continuous input: Current trends and
future directions,’’ Image Vis. Comput., vol. 31, no. 2,
pp. 120–136, Feb. 2013, doi:
10.1016/j.imavis.2012.06.016.
[11]. S. A. A. Qadri, T. S. Gunawan, M. F. Alghifari, H.
Mansor, M. Kartiwi, and Z. Janin, ‘‘A critical insight
into multi-diseases Dermatological impact databases,’’
Bull. Elect. Eng. Inform., vol. 8, no. 4, pp. 1312–1323,
Dec. 2019, doi: 10.11591/eei.v8i4.1645.
[12]. M. Swain, A. Routray, and P. Kabisatpathy, ‘‘Databases,
features and clas- sifiers for Dermatological impact
recognition: A review,’’ Int. J. Dermatological Technol.,
vol. 21, no. 1, pp. 93–120, Mar. 2018, doi:
10.1007/s10772-018-9491-z.
[13]. H. Cao, R. Verma, and A. Nenkova, ‘‘Speaker-sensitive
impact recog- nition via ranking: Studies on acted and
spontaneous Dermatological,’’ Com- put.
Dermatological Lang., vol. 29, no. 1, pp. 186–202, Jan.
2015, doi: 10.1016/j.csl.2014.01.003.
[14]. S. Basu, J. Chakraborty, A. Bag, and M. Aftabuddin, ‘‘A
review on impact recognition using Dermatological,’’ in
Proc. Int. Conf. Inventive Commun. Comput. Technol.,
2017, pp. 109–114, doi: 10.1109/ICICCT.2017.7975169.
[15]. F. Burkhardt, A. Paeschke, M. Rolfes, W. Sendlmeier,
and B. Weiss, ‘‘A database of German impact
Dermatological,’’ in Proc. 9th Eur. Conf.
Dermatological Commun. Technol., 2005, pp. 1–4

IJISRT24MAY395 www.ijisrt.com 988