A Thorough review on Med Bot using Deep Learning

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 04 | Apr 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 73
A Thorough review on Med Bot using Deep Learning
Mahesh Kamble, Guided by Prof. Nagaraju Bogiri
Mahesh Kamble, Dept of Computer Engineering, K J College of Engineering and Management Research,
Maharashtra, India
Prof. Nagaraju Bogiri, Dept of Computer Engineering, K J College of Engineering and Management Research,
Maharashtra, India
---------------------------------------------------------------------***--------------------------------------------------------------------
Abstract - Good heath is one of the most important and an
essential aspect of any individual’s life. Being healthy and fit is
an amazing experience and can improve the quality of life ofa
person considerably. Health is most of the times equated to
wealth which is a very apt comparisonashavingagoodhealth
can have a myriad range of benefits that are irreplaceable.
The health allows the individual to achieve increased
productivity and the peace of mind that is difficult to achieve
in a compromised health. With the recent pandemic and the
rise in the number of individuals that are suffering from
various illnesses which can be attributedtothelifestyles which
have become hectic and highly stressful. This makes the
individuals highly susceptible to infections and other diseases
which can detrimental to their overall well being. Thereisalso
a high demand for achieving remote diagnosis which can be
helpful for the doctor as well as the patients. Therefore, the
previous approaches on this topic have analyzed effectively
implement the paradigm of Disease predictionandsuggestion
through the realization of a medical Chatbot which will be
detailed in the next set of research papers on this paradigm.
Key Words: Pearson Correlation, k Nearest Neighbors,
Linear Regression, Hidden Markov Model and Decision
Tree.
1.INTRODUCTION
A good health is extremely vital to the overall well
being and the improvement in day to day activities andtasks
that are performed by that individual. Being healthy allows
the person to work at their maximum potential which can
lead to a much satisfying experience. The maintenance of
health is a complex and a consciousendeavorwhichrequires
an individual to lead a very healthy lifestyle. But due to the
activities of the modern world, there are a considerable
section of the population that do not lead a very healthy life
and have a lifestyle that does not allow for an effective
maintenance of the health. This has led to a considerable
increase in the illnesses and other ailments that could have
been prevented by leading a better lifestyle.
The increase in the illnesses have been noticed
across the world with the development of the cities and the
lack of work and life balance. This kind of lifestyle leads to a
lot of stress and the lack of care for one’s health. Health care
system is among the motivating factors that might impact a
person's ability to make effective use of certain available
resources to them. Medical management might be
tremendously advantageous in a multitude of areas,
particularly effective technological developments and a
significant rise in living standards of the people. As a
consequence, medical organizations and the bio medicine
industry have been recognized as among the field's biggest
important and important themes. Large-scale innovations
have primarily focused on medical breakthroughs that have
been shown to be effective in extending and enhancing a
human's life expectancy. Thishasonlybeenaccomplishedby
routinely preserving human health and removingextremely
damaging ailments and serious disorders.
The advancements are and have been immensely
valuable in maintaining human health and wellness
potentially fatal disorders. Scientists may now facilitate the
learning and collaborate on countless novel therapies and
other precautionary measures in a relatively short time
thanks to the emergence of the digital service, which has
made significant contributions to this field of study. The
notion that the World Wide Web has aroused an essentially
ubiquitous response from customers, with the majority of
the public being connected to this global communications
architecture, has broad implications. This link has fostered
increasing involvement amongst individuals since the
internet infrastructure can be used for communicating
effectively. As a response, a number of chatbots and other
digital medical assistance websites have been created to aid
the wider population with their health complications.
With the massive amount of people suffering from
chronic ailments all over the globe, the medical system is
under mounting pressure to provideappropriatehealthcare
to this growing population. Medical providers have been put
under a lot of pressure to help these people be diagnosed
and treated as quickly as possible. Physicians are under a
great deal of stress due to their heavy schedule, which can
result in a lot of human mistakes and other issues. Add to
that the reality that almost all individuals with physical
challenges are unable to travel hundredsofmilestoseetheir
doctor for a definitive reading andregular exams.Asa result,
this technique has been useful incomprehendingpastefforts
for medical chatbot deployments, as well as in attaining our
strategy, which will be detailed in future installments of this
research.

This literature surveypaper segregatesthesection2for
the evaluation of the past work in the configuration of a
literature survey, and finally, section 3 provides the
conclusion and the future work.
2. RELATED WORKS
I. Ghafir et al. [1] presented BotDet, a revolutionary
technique for detecting botnet C&C traffic. The created
system (BotDet) is divided into two phases,thefirstofwhich
contains developed modules for detecting suspected botnet
C&C communication strategies. The second step employs an
alert correlation mechanism based on voting among
detecting modules. BotDet has an 82.3 % detection rate and
a 13.6 % false alarm rate, respectively. Blacklist automation,
which is utilized in some detection modules, depends on
several intelligence inputs. This enables BotDet to identify
attacks in real-time. BotDet balances the true positive and
false positive rates with 82.3 % and 13.6 %, respectively,
according to the assessment findings.
X. Ren et al. proposed an experimental project that
used conversational agent-based interactions to enhance
intelligent decision assistance during healthcare
consultations. ConsultAI, an interactive chatbot assistant,
was utilized to implement the proposed conversational
method, which was designed to give real-time help to the
occupational health physician [2]. The authors conducted
field research witheightoccupational healthconsultationsto
determine the practicality of ConsultAI in the context of
occupational health; and the impact of ConsultAI chatbot
interaction styles on the user experience. The quantitative
results demonstrate that physicians rated ConsultAI's
conversational interface highly in terms of information
trustworthiness and technological adoption.
R. Rajkumar et al. investigated the links among
Introvert and Extravert personality types, as well as their
ways of learning [3]. Initially, using modified VARK surveys,
a Chatbot is often used to categorize people as Introverts or
Extroverts. The Chatbot participants' responses have been
shown to produce high-quality data. The learners are given
two minutes of visual and aural content to view in a calm
environment, based on the Chatbot's classifications. While
learners are seeing the information, their Beta brain waves
are caught and a dataset is produced in a one-second time.
This information is verified utilizing machine learning
classification methods such as Naive Bayes, N48 tree, and
Clustering algorithms. The suggested strategyhasbeen built
to enhance the precision of learner categorization. The
suggested Bio-Inspired Chatbot requires less time than
existing approaches to classify learners.
For multi-turn response selection in retrieval-
depend chatbots, G. Mao et al. suggest a hierarchical
aggregation network of multi-presentation. The authors
create self-aggregationandmatchingaggregationtechniques
for hierarchical aggregation. Twotechniquesintegratemulti-
grained representations step by step, allowing for the
distillation of high-level information and the reduction of
redundancies. The authors considerthecandidateanswerto
be a legitimate part of the context, and by including it into
the model framework, they hope to enhance it. Experiments
on two large-scale response selection data sets reveal that
their methodoutperformscurrent best-practicemethods [4].
The authors provide a visualization result to show that the
model can capture important information for response
selection. They next conduct ablation evaluations to
investigate each module's impact, and the results support
their utility and efficiency.
G. Daniel et al. [5] presented Xatkit, a multi-channel
and multiplatform chatbot modeling framework. Xatkit
offers a set of domain-specific languages for decoupling
chatbot definition from platform-specific properties. This
increases the reusability of thechatbotandmakesit easierto
redeploy it when the company's needs change, includingthe
ability to update the NLU engine used during the text
analysis phase. In addition to the actions and events
contained in the current version of Xatkit, the runtime
component may simply be updated to handle new platform-
specific actions and events. Additional contributors, for
example, have recently contributed Alexa and Trello to the
core Xatkit team.
M. Polignano et al. introduced HealthAssistantBot, a
Telegram-based conversational assistant for assisting
patients in their everyday tasks. The agent was built with a
modular strategy in mind so that new features can be added
to it as needed. Users may track their therapies, biological
parameters, receive doctor recommendations, and self-
diagnose using the system [6]. The dialogueiscarriedout via
a text-based interface, which makesitsimpletointeractwith
while also being resistant to mistakes. The interface,
gateway, and server-side operations are the three primary
aspects of the proposed platform's design. Each of them is
self-contained to ensure strong internal coherence and little
overlap with the other modules' functions.
H. Honda et al. presented deep learning strategies
for learning symbolic processingandtrainedmodel methods
for building question answering systems. The suggested
approaches have rich representations and great resilience,
and these models can learn even from small-scale data,
according to experimental resultsonthetrainingofsymbolic
processing models. The capacity to handle unknown input,
especially when employing Word2Vec, will be a significant
addition to artificial intelligence research [7]. Furthermore,
the experimental findingsof thequestionansweringsystems
revealed that viable question answering systems might be
constructed using Prolog knowledge bases. Using a
connectionism-based strategy to build such systems will be
incredibly tough.

To increase the informativeness and fluency of the
generated answer, L. Zhang et al. integrate the generate-
then-polish procedure into the response generation. The
authors suggest a GP with two encoders, one for
representing context and the other for representing the
prototype, as well as one decoder for response creation and
polishing. A set of experimental findings on a Chinese
dialogue corpus show that their suggested model has a
significant advantage [8]. On the Doubandialoguedataset, in
particular, the proposed model produces state-of-the-art
results.
Over linked data, A. Ait-Mlouk et al. suggested a
knowledge graph-based chatbot system that is designed for
community engagement [9]. Large-scale, publicly available
knowledge bases, multilingual, speech-to-text, and external
APIs are all utilized in the proposed KBot system. KBot also
makes use of machine learning and natural language
understanding technologies, such as named entity
recognition, factoids, and repeated queries, as well as
dialogue management. ThesuggestedKBothasenhancedthe
end-to-end user experience in terms of interactive question
answering and performance, according to a usability
investigation. It is more practical for information retrieval,
acquisition, intent classification, query comprehension, and
continuous learning.
D. Carlander-Reuterfeltetal.recognizedthe benefits
of cognitive aids in education as well as the implementation
hurdles. As a consequence, they presented a chatbot named
JAICOB, the students will have access to this chatbot with a
user-friendly interface and a human-like experience. It can
deliver information and dispel doubts regarding Data
Science. The primary contribution is adapting the
architecture to the real pedagogic demands of the students
and being flexible in continuing a conversation. It may also
be used by teachers to discover gaps in their pupils'
knowledge [10]. They can also delegate the respondingofall
queries to Jaicob. The pedagogue is also a wonderful asset
for selecting the most useful sources of information for
Jaicob to feed on, resulting in a curated source of knowledge
rather than a standard Google Search. The project was
assessed by a group of students, and it received high marks
for usability and originality.
EBER, anintelligentchatbot, wasdemonstratedby S.
García-Méndez et al [11]. To their knowledge, this is thefirst
system to use AIML, NLG, and SA to create a brief,
contextualized conversations that serve as connectors
between newscasts. EBER functions genuinely as an
"intelligent radio" for amusing older people thanks to this
combination. The connections between behavioral factors
and sample demographics were well-defined, adding to the
credibility of the user satisfaction scores. The technology
enhances contentcharacterizationevenforinattentive,tired,
or confused persons by automatically collecting knowledge
from connecting interactions with a positive attitude.
T. -Y. Chen et al. offer the YMC model, a basic yet
effective method of capturing video information for the
classification of user intent. The model's main principle is to
mask off irrelevant areas depending on the object detection
result, forcing the next classifier to focus on the important
portions of the picture. In addition, the authors use
Autoencoder, an unsupervised approach, to encode multi-
modal information, such as the concatenation of textual and
visual characteristics, into smaller dimensions [12]. The
authors can not only execute faster during inference with
these smaller dimensions, but they can also maintain high
performance and accuracy. The findings suggest that using
YOLOv4 as the object detection model improves
performance marginally overusing YOLOv3 as the object
detection model.
G. A. Santos et al. presented the Chatbot
Management Process (CMP) as a technique for managing
chatbot material [13]. It consists of six processes separated
into three phases. CMP is a cyclic process that adapts to the
demands of the company and is based on real-world user
dialogues, which is its driving power. The CMP is a post-
deployment management mechanism for machine learning
chatbots. It includes steps for changing the knowledge base,
building models,testingmodifications,andanalyzingmetrics
to determine the health of the chatbot. It's easy to assign
tasks to employees with diverse skill sets with CMP, and
each team member's responsibilitiesareclearlydefined. The
EvaTalk System, a full platform that includes both the
chatbot interface and administration tools for post-
deployment maintenance, was used to validate the
technique. EvaTalk demonstrated that the CMP can scale to
meet the needs of a high-demand chatbot with the correct
tools and personnel. Also, as long as the organization's goals
are well connected with measurements, the analysis phase
proved to be quite crucial for the process.
E. H. -K. Wu et al. explore and analyze how well
existing chatbot technologiesenhanceusers'educationonE-
Learning platforms, as well as how these techniques might
be leveraged to address concerns like separation and
detachment. They devised a chatbot that acts as an E-
Learning assistant for testing purposes. The NLP foundation
of their chatbot is composed of two models: retrieval-based
and QANet [14]. This hybrid chatbot with two models was
designed to operate in an e-learning environment. The core
answer context of their chatbot isn't just for course content,
but also for general conversation and chitchat, making it
seem more like a real companion.
J. G. Nangoy et al. explain their study on a chatbot
that can handle picture messages and respond with product
information, and they propose a method for categorizing
pictures on the LINE @ platform's chatbots [15]. Thesystem
is designed to respond to buyer questions concerning a
seller's product specifications. The administrator control
panel is where product informationiscollected. Thetest was

carried out on four different CNN models with different
layers and epochs. Model A has three layers totaling 300
epochs, model B has two layers totaling 40 epochs, model C
has two levels totaling 100 epochs, and model D has two
layers totaling 300 epochs. Dependingonthe resultsofthese
tests, the C model has the highest accuracy value when
compared to the other two models.
3. CONCLUSIONS AND FUTURE SCOPE
The fast paced lifestyle has become one of the most
challenging and problematic occurrences which have been
affecting the populations and their health. The individuals
have very less time to care for their health which can lead to
a lot of problems. The paradigm of remote health
management can come to the rescue in this regard. Medical
chatbots can also be used to accurately diagnose illnesses
related health issues reported by a patient via the internet.
As a result, there is a rising demand for a remote diagnostic
method that may considerably benefit theoverall healthcare
industry. Through the use of machine learning techniques,
this autonomous chatbot can assess complaints and deliver
an early diagnosis. This can cut rehabilitation time in half
and offer the client with therapy in a relatively short time. A
number of traditional research have been discovered and
investigated in depth for the aim of completing thisresearch
survey paper. The approach achieved through this analysis
has been realized which will be elaborated in thenextarticle
on this topic.
REFERENCES
[1] I. Ghafir et al., "BotDet: A System for Real-Time Botnet
Command and Control Traffic Detection," inIEEEAccess,vol.
6, pp. 38947-38958, 2018, DOI:
10.1109/ACCESS.2018.2846740.
[2] X. Ren, G. Spina, S. De Vries, A. Bijkerk, B. Faber and A.
Geraedts, "Understanding Physician’s Experience With
Conversational Interfaces During Occupational Health
Consultation," in IEEE Access, vol. 8, pp. 119158-119169,
2020, DOI: 10.1109/ACCESS.2020.3005733.
[3] R. Rajkumar and V. Ganapathy, "Bio-Inspiring Learning
Style Chatbot Inventory Using Brain Computing Interface to
Increase the Efficiency of E-Learning," in IEEE Access, vol. 8,
pp. 67377-67395, 2020, DOI:
10.1109/ACCESS.2020.2984591.
[4] G. Mao, J. Su, S. Yu, and D. Luo, "Multi-Turn Response
Selection for Chatbots With Hierarchical Aggregation
Network of Multi-Representation," in IEEE Access, vol.7,pp.
111736-111745, 2019, DOI:
10.1109/ACCESS.2019.2934149.
[5] G. Daniel, J. Cabot, L. Deruelle and M. Derras, "Xatkit: A
Multimodal Low-CodeChatbotDevelopmentFramework,"in
IEEE Access, vol. 8, pp. 15332-15346, 2020, DOI:
10.1109/ACCESS.2020.2966919.
[6] M. Polignano, F. Narducci, A. Iovine, C. Musto, M. De
Gemmis and G. Semeraro, "HealthAssistantBot: A Personal
Health Assistant for the Italian Language," in IEEE Access,
vol. 8, pp. 107479-107497, 2020, DOI:
10.1109/ACCESS.2020.3000815.
[7] H. Honda and M.Hagiwara,"QuestionAnsweringSystems
With Deep Learning-Based Symbolic Processing," in IEEE
Access, vol. 7, pp. 152368-152378, 2019, DOI:
10.1109/ACCESS.2019.2948081.
[8] L. Zhang, Y. Yang, J. Zhou, C. Chen and L. He, "Retrieval-
Polished Response Generation for Chatbot," in IEEE Access,
vol. 8, pp. 123882-123890, 2020, DOI:
10.1109/ACCESS.2020.3004152.
[9] A. Ait-Mlouk and L. Jiang, "KBot: A Knowledge Graph-
Based ChatBot for Natural Language Understanding Over
Linked Data," in IEEE Access, vol. 8, pp. 149220-149230,
2020, DOI: 10.1109/ACCESS.2020.3016142.
[10] D. Carlander-Reuterfelt, Á. Carrera, C. A. Iglesias, Ó.
Araque, J. F. Sánchez Rada and S. Muñoz, "JAICOB: A Data
Science Chatbot," in IEEE Access, vol. 8, pp. 180672-180680,
2020, DOI: 10.1109/ACCESS.2020.3024795.
[11] S. García-Méndez, F. De Arriba-Pérez, F. J. González-
Castaño, J. A. Regueiro-Janeiro and F. Gil-Castiñeira,
"Entertainment Chatbot for the Digital Inclusion of Elderly
People Without AbstractionCapabilities,"inIEEEAccess,vol.
9, pp. 75878-75891, 2021, DOI:
10.1109/ACCESS.2021.3080837.
[12] T. -Y. Chen, Y. -C. Chiu, N. Bi and R. T. -H. Tsai, "Multi-
Modal Chatbot in Intelligent Manufacturing," inIEEEAccess,
vol. 9, pp. 82118-82129, 2021, DOI:
10.1109/ACCESS.2021.3083518.
[13] G. A. Santos, G. G. de Andrade, G. R. S. Silva, F. C. M.
Duarte, J. P. J. D. Costa, and R. T. de Sousa, "A Conversation-
Driven Approach for Chatbot Management," in IEEE Access,
vol. 10, pp. 8474-8486, 2022, DOI:
10.1109/ACCESS.2022.3143323.
[14] E. H. -K. Wu, C. -H. Lin, Y. -Y. Ou, C. -Z. Liu, W. -K. Wang
and C. -Y. Chao, "Advantages and Constraints of a Hybrid
Model K-12 E-Learning Assistant Chatbot," in IEEE Access,
vol. 8, pp. 77788-77801, 2020, DOI:
10.1109/ACCESS.2020.2988252.
[15] J. G. Nangoy and N. H. Shabrina, "Analysis of Chatbot-
Based Image Classification on Social Commerce LINE@
Platform," 2020 7th NAFOSTED Conference on Information
and Computer Science (NICS), 2020, pp. 232-237, DOI:
10.1109/NICS51282.2020.9335874.

A Thorough review on Med Bot using Deep Learning

More Related Content

Similar to A Thorough review on Med Bot using Deep Learning (20)

More from IRJET Journal (20)

Recently uploaded (20)

A Thorough review on Med Bot using Deep Learning