2023 6th International Conference on Contemporary Computing and Informatics (IC3I)

Multiple Disease Prediction Using ML and Doctor

Recommendation by Sentiment Analysis
Dr. Kakoli Banerjee Vinooth P
Department of Computer Science and Engineering Department of Computer Science and Engineering
JSS Academy of Technical Education, JSS Academy of Technical Education
Noida, India. Noida, India.
[email protected]

Abstract - For the purpose of preventing and treating occurring diseases in the early phase as when they are not
illness, accurate and prompt examination of any health- checked or examined they can turn into a disease and
related issue is crucial. A dangerous illness might not be more dangerous diseases can even cause death. This
properly diagnosed using the conventional methods. The system will predict the most possible disease based on the
development of a machine learning (ML)-based medical
2023 6th International Conference on Contemporary Computing and Informatics (IC3I) | 979-8-3503-0448-0/23/$31.00 ©2023 IEEE | DOI: 10.1109/IC3I59117.2023.10397715

given symptoms by the user and recommend doctors will

diagnosis system for disease prediction can lead to a
diagnosis that is more precise than one made using be based on the sentiment analysis of patients reviews.
traditional techniques.
II. LITERATURE REVIEW
Using various ML algorithms, we have created a system
Numerous studies have been conducted on the topic
for disease prediction. More than 50 diseases were present in
the data set that was processed. The diagnosis system of disease prediction utilizing various machine learning
provides the output as the disease that an individual may be approaches and algorithms that can be employed by
experiencing based on the symptoms, age, and gender of the medical organizations. There are also various studies
individual. In order to ensure that treatment can begin on regarding sentiment analysis of reviews .This essay
time and lives can be spared, our diagnosis model can serve examines a few of those studies from research
as an assistance for doctor by the early diagnosis of a disease publications using their methods and findings. Following
and also recommend the nearby best possible doctor for the are reviews:
predicted disease to the user. The recommendation of
doctors will be based on the sentiment analysis of patients
In their research, MIN CHEN et al. [1] used machine
reviews.
learning methods to create an disease prediction system.
Keywords: Disease Prediction,Machine He applied methods such as the CNN-UDRP algorithm,
Learning,Symptoms,Sentiment Analysis CNN-MDRP algorithm, Naive Bayes, K-Nearest
Neighbor, and Decision Tree to the prediction of disease.
I. INTRODUCTION The precision of this suggested system was 94.8.

The field of machine learning uses historical data to Disease Risk Prediction was advised by Sayali
make predictions. The concept of a computer system Ambekar et al., [2] who carried out the task using a
known as "machine learning" refers to how a machine convolution neural network. Machine learning methods
learning model learns from data and experience. including the CNN-UDRP algorithm, Naive Bayes, and
KNN algorithm are employed in this research. The system
There are two stages to the machine learning employs structured data to be trained, and Naive Bayes is
algorithm: 1) Testing and 2) Training. used to attain an accuracy of 82%.

In order to improve processes and give patients better Using a fuzzy approach, Naganna Chetty et al. [3]
care, the healthcare industry also uses machine learning. built a system that provides better outcomes for disease
The disease prediction system forecasts illnesses based on prediction.He used KNN classifier, fuzzy c-means
the patient's symptoms and by using sentiment analysis clustering, and fuzzy KNN classifier approaches. The
recommended doctors for predicted ailment. accuracy of the diabetic disease and liver disorder
predictions in this research is 97.02% and 96.13
Sentiment analysis has developed into a potent tool respectively.
for tracking and comprehending online reviews as
people express their opinions and feelings about The focus of the Senthilkumar Mohan et al study [4] was
something more honestly than ever before. So, by hybrid machine learning approaches, which employ
applying sentiment analysis on the doctor reviews by algorithms such as Decision Tree, Support Vector
patients gives the best doctor for particular disease along Machine, Random Forest, Naive Bayes, Neural Network,
with other factors. and KNN to effectively anticipate cardiac illness. This
system's accuracy rating is 88.47%.
The systems that are now on the market are either
specifically designed to treat a specific condition, are in Data mining for the prediction of diabetic disease was
development, or are being researched to provide a solution examined by Deeraj Shetty et al. [5] using Naive Bayes
to the problem of generalized disease. The main motive of and KNN algorithms. The accuracy of this system's
the proposed system is the prediction of the commonly diabetes prediction is higher than Naive Bayes thanks to

1469
979-8-3503-0448-0/22/$31.00 ©2023 IEEE
Authorized licensed use limited to: Don Bosco Institute of Technology-Bengaluru. Downloaded on December 19,2024 at 06:51:52 UTC from IEEE Xplore. Restrictions apply.
 2023 6th International Conference on Contemporary Computing and Informatics (IC3I)

KNN. extraction was a useful technique for deciphering online

customer suppositions. An organisation can improve its
Pahulpreet Singh Kohli et al. [6] state that machine marketing efforts thanks to the data obtained from online
learning methods and tools like logistic regression, platforms and product review websites. The customer
decision trees, support vector machines, random forests, purchase selections are influenced and informed by the
and adaptive boosting can be used to predict diseases. The product reviews.
primary focus of this essay is the prediction of diabetes,
breast cancer, and heart disease. Logistic regression yields A probabilistic generative model was created by
the highest accuracy rates: 95.71% for diabetes, 84.42% Wallace et al. [14] to collect latent sentiment across care-
for heart disease, and 87.12% for breast cancer. related features. They demonstrated that adding the
model's output to regression models enhances correlations
Utilizing distributed machine learning classifiers, with state-level quality metrics.
Lambodar Jena et al. [7] concentrated on risk prediction
for chronic diseases using methods like Naive Bayes and Using topic modelling, Hao and Zhang [15] were able
Multilayer Perceptron. The accuracy of Naive Bayes and to identify themes that were common to four different
Multilayer Perceptron in this paper's attempt to forecast specialties in the doctor reviews they had gathered from
Chronic Kidney Disease is 95% and 99.7%, respectively. Good Doctor Online. In all four disciplines, they
discovered four issues that were frequently discussed: the
Dhomse Kanchan B. et al. [8] studied the prediction of process of locating doctors, technical prowess or bedside
certain illnesses using principal component analysis and manner, patients' praise of the doctor, and symptom
machine learning techniques, such as Naive Bayes descriptions. Similar to this, Hao et al. employed topic
classification, Decision Tree, and Support Vector Machine modelling to compare reviews across RateMDs and Good
approaches. For diabetes, this method is 34.89% accurate, Doctor Online, two websites that allow patients to rate and
while for heart disease, it is 53% accurate. evaluate US doctors.

Ankita Dewan et al[9] .'s recommendation Using a 4-step method, Hu and Liu [16] retrieved
of prediction system of disease uses a hybrid data mining opinions of features from customer reviews. To detect
classification technique. This system employs methods features, this algorithm uses association rule mining. It
like Naive Bayes, Decision Trees, and Neural Networks. then prunes irrelevant and redundant features, finds rare
This system has an accuracy rate of 87%. features, and lastly determines the semantic orientation of
each opinion sentence.
Anjan Nikhil Repaka et al. [10] used naive Bayesian to
construct and implement a prediction model for heart Several manually created criteria were implemented
disease. Any user may access the forecast results by by Agarwal et al. [17] in order to derive dependency tree
utilizing this method with any smartphone device. This patterns from phrases. Combining this data with the
method has an accuracy of 89.77%. semantic information from the Massachusetts Institute of
Technology Media Lab ConceptNet ontology, they trained
Gao et al. [11] examined patterns in physician evaluations a machine learning model to identify concept patterns in
over time to determine the factors influencing Web-based the text by using the concepts that were extracted. This
ratings. They found that, in general, evaluations were allowed them to classify documents into positive and
positive and that doctors who had been practicing for a negative categories.
long period, such as obstetricians or gynecologists, were Text from electronic medical records was used by Lix
more likely to obtain reviews than other medical et al. [18] to apply an SVM classifier to identify patients
specialties. They also found that clinicians without who had taken alcohol. A bag-of-words model was used
malpractice claims, board-certified physicians, highly to represent unigrams and bigrams in these data.
respected medical school grads, and recent graduates all
had better ratings. YeongWai Chung [19] Conduct research on huge
data and provide sentiment analysis problems. Twitter
Jiugang Li et al. [12] developed a hashtag produces 175 million tweets every day on average. 1
recommender system using the skip-gram model and Zettabyte of data has already been produced by the globe.
convolutional neural networks (CNN) to learn semantic
phrase vectors, taking into account the importance of III. CURRENT ISSUES
hashtags in sentiment analysis. These vectors employ
LSTM RNN to classify hashtags based on the features. Traditional diagnostic techniques involve physically
Results show that this model outperforms more widely assessing patients to measure things like body
used models like SVM and Standard RNN. This temperature, pulse rate, heartbeat, blood pressure, and
investigation is based on the fact that it was subjected to asking about their medical history. There are various
standard AI approaches like SVM and collaborative limitations of these techniques of diagnosis –
filtering; the semantic features are lost, which has a
significant impact on obtaining a reasonable expectation. ● It doesn't reveal the reason of the patient's illness.
● Prolonged procedure of diagnosis.
Jain, Kumar, and Mahanti [13] found that sentiment ● Don't give enough information on the illness.

1470

Authorized licensed use limited to: Don Bosco Institute of Technology-Bengaluru. Downloaded on December 19,2024 at 06:51:52 UTC from IEEE Xplore. Restrictions apply.
 2023 6th International Conference on Contemporary Computing and Informatics (IC3I)

● Increase the risk of developing a disease in the sarcasm, irony, humour, which a person has minimal
future trouble recognising.
In reality, sentiment analysis is a difficult task even for
Correct doctor for specific disease is mandatory for humans, so sentiment analysis classifiers might not be as
curing of the disease. Due to lack of knowledge about the accurate as other classifiers.
doctor, people may be misdiagnosis or taking more than
enough time to recover. So, recommendation of proper
doctor is must. VI. FUNDING

IV. FUTURE SCOPE This research received no external funding.

According to the information and symptoms users VII. CONFLICT OF INTEREST

provide to the web-based application, this prediction
system can be utilized to provide timely advice on their There are no conflicts of interest, according to the
ailment. To determine the disease that would be most authors.
closely associated to the patient's details, several clever
data processing techniques are applied in this case. The REFERENCES
patient can then get in touch with the appropriate disease
[1] M. Chen, Y. Hao, K. Hwang, L. Wang, and L. Wang, “Disease
specialist recommending by doctor recommend model of
prediction by machine learning over big data from healthcare
application and undergoes treatment based on the test communities” IEEE Access, vol. 5, no. 1, pp. 8869–8879, 2017.
results. You can utilize this method to get a free [2] Sayali Ambekar, Rashmi Phalnikar, “Disease Risk Prediction by
consultation on any illness. Additionally, it eliminates the Using Convolutional Neural Network” IEEE, 978-1-5386-5257-
2/18, 2018.
need for an initial general physician visit. The patient can
[3] Naganna Chetty, Kunwar Singh Vaisla and Nagamma Patil, “An
receive a prognosis and direct appointment from a doctor Improved Method for Disease Prediction using Fuzzy Approach”
who specialists in that particular subject if they want to do IEEE, DOI 10.1109/ICACCE.2015.67, pp. 569-572, 2015.
so. [4] Dhiraj Dahiwade, Gajanan Patle and Ektaa Meshram, “Designing
Disease Prediction Model Using Machine Learning Approach”
IEEE Xplore Part Number: CFP19K25-ART; ISBN: 978-1-5386-
V. CONCLUSION 7808-4, pp. 1211-1215, 2019.
[5] Lambodar Jena and Ramakrushna Swain, “Chronic Disease Risk
Everyone's daily life are impacted by disease Prediction using Distributed Machine Learning Classifiers” IEEE,
978-1-5386-2924-6/17, pp. 170-173, 2017. Cai, C.W., 2018.
prediction using machine learning, but individuals
Disruption of financial intermediation by FinTech: a review on
working in the healthcare sector utilize these systems crowdfunding and blockchain. Accounting & Finance, 58(4),
frequently to forecast patients' illnesses based on their pp.965-992.
demographics and symptoms. [6] Pahulpreet Singh Kohli and Shriya Arora, “Application of Machine
Learning in Disease Prediction” IEEE, 978-1-5386-6947-1/18, pp.
1-4, 2018.
On average, a prediction accuracy probability of 95% [7] Deeraj Shetty, Kishor Rit, Sohail Shaikh and Nikita Patil, ”
is attained. Our diagnosis model can serve as an assistance Diabetes Disease Prediction Using Data Mining” IEEE, 978-1-
for doctor by the early diagnosis of a disease and also 5090-3294-5/17, 2017.
[8] Ankita Dewan and Meghna Sharma, “Prediction of Heart Disease
recommend the nearby best possible doctor for the
Using a Hybrid Technique in Data Mining Classification” IEEE,
predicted disease. The recommendation of doctors will be 978-9-3805-4416-8/15, pp. 704-706, 2015
based on the sentiment analysis of patients reviews. [9] Senthilkumar Mohan, Chandrasegar Thirumalai and Gautam
Srivastava, “Effective Heart Disease Prediction Using Hybrid
Machine Learning Techniques” IEEE Access, DOI
Early diagnosis of diseases can both lengthen your
10.1109/ACCESS.2019.2923707, pp. 81542-81554, 2019.
life and spare you from financial hardship. We have [10] Anjan Nikhil Repaka, Sai Deepak Ravikanti and Ramya G
utilized a variety of machine learning algorithms, Franklin,”Design And Implementing Heart Disease Prediction
including to get the highest level of accuracy, use Random Using Naives Bayesian” IEEE Xplore Part Number: CFP19J32-
ART; ISBN: 978-1-5386-9439-8, pp. 292-297, 2019.
Forest and K nearest neighbor (KNN).
[11] P Kumar, T Choudhury, S Rawat, S Jayaraman ,Analysis of
various machine learning algorithms for enhanced opinion mining
Doctor Recommendation module is a model to using Twitter data streams, International Conference on Micro-
anticipate the comment's mood from the review content. Electronics, 2016
[12] Smith R, Lipoff J. Evaluation of dermatology practice online
SMOTE is employed to remedy the issue because of the
reviews: lessons from qualitative analysis. JAMA Dermatol. 2016
dataset's unbalanced distribution. Feb;152(2):153–157. doi: 10.1001/jamadermatol.2015.3950
[13] Wallace BC, Paul MJ, Sarkar U, Trikalinos TA, Dredze M. A
Despite being text data of patient’s reviews, the large-scale quantitative analysis of latent factors and sentiment in
online doctor reviews. J Am Med Inform Assoc. 2014 Jun
distribution, characteristics, and choice of machine
10;21(6):1098–1103. doi: 10.1136/amiajnl-2014-002711.
algorithm all have an impact on how accurately machine [14] Hao H, Zhang K. The voice of Chinese health consumers: a text
learning predictions turn out. Word clouds can be used to mining approach to web-based physician reviews. J Med Internet
examine the words that occur frequently. Res. 2016 May 10;18(5):e108. doi: 10.2196/jmir.4430.
[15] Matsumoto S, Takamura H, Okumura M. Sentiment classification
using word sub-sequences and dependency sub-trees. 9th Pacific-
The module's drawback is the sentiment analysis Asia Conference on Advances in Knowledge Discovery and Data
forecast accuracy, which is only about 70-80% accurate. Mining (PAKDD'05); May 18-20, 2005; Hanoi, Vietnam. Berlin:
In contrast, computer algorithms struggle to recognise Springer; 2005. pp. 301–311.

1471

Authorized licensed use limited to: Don Bosco Institute of Technology-Bengaluru. Downloaded on December 19,2024 at 06:51:52 UTC from IEEE Xplore. Restrictions apply.
 2023 6th International Conference on Contemporary Computing and Informatics (IC3I)

[16] Yadav, N., Banerjee, K., & Bali, V. (2020). A survey on fatigue
detection of workers using machine learning. International Journal
of E-Health and Medical Communications (IJEHMC), 11(3), 1-8
[17] Banerjee, K., & Bali, V. (2020). Design and Development of
Bioinformatics Feature Based DNA Sequence Data Compression
Algorithm. EAI Endorsed Trans. Pervasive Health Technol., 5(20),
e5.
[18] Banerjee, K., Kumar, M. S., & Tilak, L. N. (2021). Delineation of
potential groundwater zones using Analytical hierarchy process
(AHP) for Gautham Buddh Nagar District, Uttar Pradesh,
India. Materials Today: Proceedings, 44, 4976-4983.
[19] Banerjee K, Bali V, Nawaz N, Bali S, Mathur S, Mishra RK, Rani
S. A Machine-Learning Approach for Prediction of Water
Contamination Using Latitude, Longitude, and Elevation. Water.
2022; 14(5):728. https://doi.org/10.3390/w14050728
[20] Banerjee K., Santhosh Kumar M.B., Tilak L.N., Vashistha S.
(2021) Analysis of Groundwater Quality Using GIS-Based Water
Quality Index in Noida, Gautam Buddh Nagar, Uttar Pradesh (UP),
India. In: Choudhary A., Agrawal A.P., Logeswaran R., Unhelkar
B. (eds) Applications of Artificial Intelligence and Machine
Learning. Lecture Notes in Electrical Engineering, vol 778.
Springer, Singapore.
[21] Sharma, T., Banerjee, K., Mathur, S., & Bali, V. (2020). Stress
analysis using machine learning techniques. International Journal
of Advanced Science and Technology, 29(3), 14654-14665.
[22] K. Banerjee and R. A. Prasad, "Reference based inter
chromosomal similarity based DNA sequence compression
algorithm," 2017 International Conference on Computing,
Communication and Automation (ICCCA), 2017, pp. 234-238,
doi: 10.1109/CCAA.2017.8229806.
[23] Banerjee, K., & Prasad, R. A. (2014, October). A new technique in
reference based DNA sequence compression algorithm: Enabling
partial decompression. In AIP Conference Proceedings (Vol. 1618,
No. 1, pp. 799-802). American Institute of Physics.
[24] K. Banerjee et al., "A review on Artificial Intelligence based Sign
Language Recognition Techniques," 2022 5th International
Conference on Contemporary Computing and Informatics (IC3I),
Uttar Pradesh, India, 2022, pp. 2195-2201, doi:
10.1109/IC3I56241.2022.10073000.
[25] K. Banerjee et al., "Prediction of Criminal Behavior Based on
Genetic Data - A Review," 2022 5th International Conference on
Contemporary Computing and Informatics (IC3I), Uttar Pradesh,
India, 2022, pp. 2206-2210, doi:
10.1109/IC3I56241.2022.10073062.
[26] K. Banerjee et al., "Assessing Water Quality Index Near Industrial
Regions and Aiding in Effective Water Management and
Controlling Water Pollution Level," 2022 5th International
Conference on Contemporary Computing and Informatics (IC3I),
Uttar Pradesh, India, 2022, pp. 1987-1991, doi:
10.1109/IC3I56241.2022.10073296.

1472

Authorized licensed use limited to: Don Bosco Institute of Technology-Bengaluru. Downloaded on December 19,2024 at 06:51:52 UTC from IEEE Xplore. Restrictions apply.