Health Checker :A Machine Learning-Powered Medical Diagnosis

Assistant
Dr.B.DWARAKANATH 1, TOM SHAJU 2,ARPAN DIXIT 3, RITAM BASU4 ,
Associate Professor1, UG Student2,3,4, Department of Information Technology SRM Institute
of Science and Technology, Ramapuram, Chennai, [email protected]
,[email protected] ,[email protected] ,[email protected].
ABSTRACT multifaceted approach enables the
system to derive real-time insights and
"Health Checker: A Machine Learning- predictions, significantly enhancing the
Powered Medical Diagnosis Assistant" diagnostic process. The objective is
delineates a cutting-edge system clear: to elevate diagnostic accuracy,
designed to revolutionize medical empower
diagnostics. Health Checker integrates
machine learning algorithms to analyze
healthcare practitioners in their
extensive patient data, including
decision-making, and ultimately
medical records and symptoms, offering
optimize patient care
real-time insights and predictions. The
through the delivery of personalized
primary goal is to improve diagnostic
treatment recommendations.
accuracy, aid healthcare professionals in
The machine learning algorithms
decision-making, and optimize patient
embedded within Health Checker
care through personalized treatment
continuously learn and adapt, evolving
recommendations. This innovative tool
with each new piece of patient data.
signifies a significant leap forward in
This adaptive learning process ensures
healthcare technology, streamlining
that the system stays at the forefront of
diagnosis processes and ultimately
medical knowledge and consistently
enhancing patient outcomes.
improves its predictive capabilities over
At its core, Health Checker is designed
time. By amalgamating historical patient
to seamlessly integrate into the existing
data, clinical expertise, and the latest
healthcare infrastructure, utilizing
medical research, Health Checker
machine learning algorithms to
becomes a dynamic and invaluable tool
meticulously analyze vast datasets
for healthcare professionals.
encompassing patient medical records
Additionally, the system serves as a
and reported symptoms. This
comprehensive decision support tool,
49
aiding clinicians in navigating the “Health Guard: Machine Learning
complexity of diagnostic scenarios and Powered Preliminary Diseases
facilitating well-informed treatment Predicton: Nisha Wilvicta J, Jitendra
decisions. Prajapati, M.A. Qhizar, MD Shadab
Arshad, Mohd. Fardeen Shaiekh”
KEYWORDS: Machine Learning(ML), (2024) [1] Healthcare is a critical
Predictive Analytics, Health domain where technological
Informatics, Electronic Health Records advancements can significantly impact
(EHR), Decision Support System. the well-being of individuals. Timely
diagnosis and treatment are fundamental
to managing various diseases
1.INTRODUCTION effectively. The "Health Guard:
Machine Learning Powered Preliminary
"HealthChecker: A Machine Learning-
Diseases Prediction" project is a novel
Powered Medical Diagnosis Assistant"
application that harnesses the power of
introduces a ground breaking system
machine learning to assist in disease
that uses machine learning in medical
prediction, making healthcare more
diagnostics. HealthChecker is a cutting-
accessible and convenient. The primary
edge tool designed to analyse intricate
objective of this project is to provide
patient data, such as medical histories
users with a simple and efficient means
and symptoms, using advanced
of predicting diseases based on their
algorithms. By offering immediate
reported symptoms. The project
insights and forecasts, HealthChecker
incorporates three machine learning
aims to refine diagnostic precision, aid
models: Decision Tree, Random Forest,
healthcare professionals in decision-
and Naive Bayes. These models are
making, and elevate patient care through
trained on a dataset that contains a list
customized treatment strategies. This
of symptoms and their associated
introduction emphasizes the significant
diseases. By utilizing this dataset, the
transformation brought about by this
models learn to recognize patterns and
technology in redefining diagnostics and
associations between symptoms and
enhancing healthcare results.
diseases, enabling them to make
informed predictions. Furthermore, the
2.RELATED WORKS
application calculates and displays the
50
accuracy of the machine learning recommenders for online merchants,
models, providing user with reliability and medication recommendations. [1]
insight. The key implementation of this goal has
been to focus on the ability to
“Suvendu Kumar Nayak, Mamata statistically estimate customers'
Garanayak, Sangram Keshari Swain, preferences for unseen objects. The
Sandeep Kumar Panda, And Deepthi objective of recommenders is typically
Godavarthi, An Intelligent Disease stated to be "assisting customers in
Prediction And Drug identifying relevant items."
Recommendation Prototype By Using 2.1 Existing works
Multiple Approaches Of Machine The current healthcare system relies on
Learning Algorithms” (2023) [2], In manual analysis of patient data and
general, a recommender prototype is a established medical expertise for
Prototype that predicts how a buyer will diagnosis. While Electronic Health
rate a certain product. Following that, Records (EHRs) have improved data
the consumer will obtain a ranking of management, the system lacks advanced
the projections. They work for several technological tools for comprehensive
well-known companies, including data analysis. There's potential for
Google, Instagram, Spotify, Amazon, enhancing diagnostic accuracy and
Reddit, and Netflix. A recommender efficiency by integrating advanced
prototype can estimate if a specific technologies like machine learning,
client will prefer an item based on their which can provide real-time insights and
profile. The recommender prototype can aid in handling complex medical data
benefit both service providers and more effectively.
consumers. They lower the transaction
costs associated with locating and 3.PROBLEM DESCRIPTION
selecting goods while making online
transactions. Current diagnostic processes face

Recommender prototypes are widely challenges in accuracy and efficiency.

used; well-known examples include Limited integration of machine learning

content recommenders for social in healthcare systems. Analysing

networking sites, , playlist generators intricate patient data, such as medical

for video and audio services, product histories and symptoms, poses a

51
considerable challenge. Lack of real- 4.SYSTEM ARCHITECTURE
time diagnostic support for healthcare
professionals. Generic treatment
approaches may result from current
diagnostic methods. Inefficiencies in
decision-making due to limited access to
immediate insights.
3.1 Proposed work

The proposed healthcare system

integrates cutting-edge technologies like
machine learning and employs boosting
algorithms to revolutionize diagnostic
capabilities. By leveraging these
advancements, it aims to overcome
Fig 4.1 SYSTEM ARCHITECTURE
existing limitations by providing
healthcare professionals with At its core, the architecture is designed to
sophisticated tools for advanced data seamlessly integrate with existing
analytics. This system facilitates real- healthcare systems and workflows. The
time insights, predictive models for architecture ensures scalability and
disease patterns, and personalized flexibility, allowing for deployment both
treatment recommendations based on on-premises and on cloud platforms.
individual patient data. The focus is on Furthermore, it prioritizes security and
enhancing efficiency, accuracy, and compliance by implementing robust
decision-making support, ultimately encryption protocols and access control
transforming diagnostic processes for mechanisms to safeguard sensitive
improved patient care outcomes. patient data. The user interface is
designed to be intuitive and user-
friendly, allowing healthcare
professionals to input patient data easily,
view diagnostic results, and interpret
model predictions with confidence.

5. SYSTEM MODULES

52
Several interconnected modules make selection, going beyond simple cleaning
up the "Health Checker: A Machine and normalization to find and keep the
Learning-Powered Medical Diagnosis most pertinent qualities that
Assistant" system, which aims to significantly aid in the diagnostic
expedite the process of medical process.
diagnosis and assistance. The system's Natural language processing
main component, a powerful Data The "Health Checker: A Machine
Collection Module, collects medical Learning-Powered Medical Diagnosis
data from many sources, such as Assistant" system's Natural Language
wearable technology, medical imaging Processing (NLP) Module is made to
systems, patient input forms, and handle unstructured textual data,
electronic health records. The Data including patient histories, medical
Preprocessing Module thoroughly notes, and research articles, in order to
preprocesses this raw data, cleaning, extract insightful information.
normalizing, and extracting pertinent Fundamentally, the NLP Module
features to get it ready for analysis. preprocesses and analyses text data
Data pre-processing using a variety of methods.
In the "Health Checker: A Machine Tokenization, which divides text into
Learning-Powered Medical Diagnosis discrete words or phrases, and
Assistant" system, the Data stemming, also known as
Preprocessing Module is an essential lemmatization, It reduces
intermediate stage that refines and dimensionality and improves
transforms raw medical data before it is consistency by standardizing words to
input into machine learning models for their root form. It can also do functions
analysis. One of its main goals is to like part-of-speech tagging, which
ensure data integrity by cleaning the recognizes the grammatical elements of
data to remove outliers, inconsistencies, a sentence, and stop-word removal,
and missing information. Additionally, it which removes common words that
normalizes information to a consistent don't have much meaning.
scale, which improves model Ensemble methods
performance and makes fair By merging several models, ensemble
comparisons. . mechanisms—such as Random Forests
This module does feature extraction and —are effective machine learning tools
53
that increase prediction accuracy and These components might include
robustness. The fundamental idea of buttons, dropdown menus, input forms,
ensemble methods is to combine the and checkboxes.
predictions of multiple different models Deployment
to generate a final forecast that The developed machine learning
frequently performs better than any one models, algorithms, and related
model used alone. A group of decision functionalities in the "Health Checker: A
trees make up the Random Forest Machine Learning-Powered Medical
Module. Each tree is trained using a Diagnosis Assistant" system are
subset of input features and a randomly transferred from a development
selected portion of training data. The environment to a production
forest's individual trees are kept diverse environment by the Deployment
in part by this unpredictability. Module so that end users can access and
Interaction and user interface use them. This module guarantees the
The "Health Checker: A Machine system's sturdiness, scalability, and
Learning-Powered Medical Diagnosis smooth integration with the medical
Assistant" system's interface and infrastructure.
interaction elements are essential for 6. RESULTS
enabling smooth communication Results and discussion offer a thorough
between users—typically medical synopsis of the conclusions and
professionals—and the system itself. ramifications of a research endeavour,
This interface acts as the entry point for experiment, or study. It entails presenting
patients to enter medical information, the findings, examining them, comparing
interact with it. them to predictions, and talking about
Simply put, the interface is designed to the importance and constraints of the
be simple to use and intuitive, making it endeavour. A smart and efficient e
appropriate for users with varying levels commerce system makes user retention a
of technical ability and medical cake walk and helps exponentially in
understanding. generating profits for a sales company.
It might have a graphical user interface Enhancing customer satisfaction, our
(GUI) with interactive components that model employs collaborative filtering
let users enter medical history, and content-based recommendation
symptoms, and other pertinent data. algorithms to precisely align each
54
product suggestion with a user's unique o Travel costs associated
tastes and preferences. with appointments and
tests.

Fig 6.1 Cost-Effectiveness

Fig 6.4 User Acceptance Chart
Comparison
 Direct Medical Costs (Blue Bar): Agreed with Health Checker (80%) -
This includes the immediate The largest slice, indicating that in 80%
expenses incurred during the of cases, clinicians agreed with the
diagnostic process, such as: diagnosis suggested by the health
o Physician consultations checker system.
o Diagnostic tests (blood
Disagreed with Health Checker (20%) -
work, X-rays, etc.)
The smaller slice, indicating that in 20%
o Initial medications or
of cases, clinicians disagreed with the
treatments
diagnosis suggested by the health
 Indirect Costs: These are the
checker system.
hidden costs associated with
traditional diagnosis, often Overall message: The pie chart suggests
overlooked: a generally high level of agreement
o Lost productivity due to between clinicians and the health
illness or doctor visits. checker system in diagnosing health
This can translate to conditions. However, there are still
missed workdays, instances (20%) where clinicians
reduced income, and a disagreed with the system's suggestions.
strain on employers.
55
 1. “Health Guard: Machine Learning
Powered Preliminary Diseases
Predicton: Nisha Wilvicta J, Jitendra
Prajapati, M.A. Qhizar, MD Shadab
Arshad, Mohd. Fardeen Shaiekh”, 2024.
2. “Suvendu Kumar Nayak, Mamata
Garanayak, Sangram Keshari Swain,
Sandeep Kumar Panda, And Deepthi
Godavarthi, An Intelligent Disease
Prediction And Drug Recommendation
Fig 6.3 Time to Diagnosis Comparison
Prototype By Using Multiple
The Health Checker app, which utilizes Approaches Of Machine Learning
machine learning, aims to streamline the Algorithms”, 2023.
diagnostic process and potentially 3. “John Adeoye, J., Koohi-Moghadam,
decrease the time it takes to reach a M., Choi, SW. et al. Predicting oral
diagnosis. cancer risk in patients with oral
leukoplakia and oral lichenoid mucositis
 Efficient Data Processing:
using machine learning.”, 2023.
Machine learning algorithms can
4. “A machine learning based data
quickly analyze large datasets of
modelling for medical diagnosis N. A.
patient information, symptoms,
Mahoto, A.
and medical records, often
Shaikh, Adel Sulaiman, Mana Saleh Al
identifying patterns that a human
Reshan, Adel Rajab, Khairan D. Rajab”,
doctor might take longer to
2023.
uncover.
5. “Lightweight Multi-Class Support
 Differential Diagnosis Support:
Vector Machine-Based Medical
The app can suggest a range of
Diagnosis System with Privacy
possible diagnoses based on the
Preservation Sherif Abdelfattah,
provided data. This aids doctors
Mohammed Baza, Mohamed Mahmoud,
by narrowing the focus,
Mostafa M. Fouda, K. Abualsaud, Elias
facilitating quicker consideration
Yaacoub, Maazen Alsabaan, Mohsen
of the most likely conditions.
Guizani”, 2023.
REFERENCES
56
6. “Virtual Medical Assistant System for System: Intelligent Data Analytics,
Diseases Detection using Machine Berlin, Germany:Springer, 2021.
Learning J. Kanimozhi, G. Preethi, N. 12. Z. Che, D. Kale, W. Li, M. T.
Mohanasuganthi, S. A. Ayshwariya, Bahadori and Y. Liu, "Deep
Lijetha. C. Jaffrin”, 2023. computational phenotyping", Proc. 21st
7. “Predicting Decision-Making Time ACM SIGKDD Int. Conf. Knowl.
for Diagnosis over NGS Cycles: An Discovery Data Mining, pp. 507-516,
Interpretable Machine Learning Aug. 2015.
Approach A. Khodabakhsh, Tobias P. 13. Z. Che, Y. Cheng, S. Zhai, Z. Sun
Loka, S. Boutin, Dennis Nurjadi, B. and Y. Liu, "Boosting deep learning risk
Renard”, 2023. prediction with generative adversarial
8. “Application of Machine Learning in networks for electronic health records",
Intelligent Medical Image Diagnosis Proc. IEEE Int. Conf. Data Mining
and Construction of Intelligent Service (ICDM), pp. 787-792, Nov. 2017.
Process Zhihong Gao, Lihua Lou, 14. Y. Cheng, F. Wang, P. Zhang and J.
Meihao Wang, Zhen Sun, Xiaodong Hu, "Risk prediction with electronic
Chen, Xiang Zhang, Zhifang Pan, health records: A deep
Haibin Hao, Yu Zhang, Shichao Quan, learning approach", Proc. SIAM Int.
Shaobo Yin, Cai Lin, Xian Shen”, 2022. Conf. Data Mining, pp. 432-440, Jun.
9. “Efficient Medical Diagnosis of 2016.
Human Heart Diseases using Machine 15. R. Miotto, F. Wang, S. Wang, X.
Learning Techniques with and without Jiang and J. T. Dudley, "Deep learning
GridSearchCV G. Ahmad, H. Fatima, for healthcare: Review opportunities
Shafi Ullah, A. S. Saidi, Asghar and challenges", Briefings Bioinf., vol.
Imdadullah”, 2022. 19, no. 1, pp. 1236-1246, Jan. 2018.
10. “C.Geetha, Dr.AR. Arunachalam, 16. P. Nguyen, T. Tran, N.
Evaluation based Approaches for Liver Wickramasinghe and S. Venkatesh,
Disease Prediction using Machine "Deepr: A convolutional net for medical
Learning Algorithms”, 2021. records", IEEE J. Biomed. Health
11. H. Malik, N. Fatema and J. A. Informat., vol. 21, no. 1, pp. 22-30, Jan.
Alzubi, AI and Machine Learning 2017.
Paradigms for Health Monitoring 17. Q. Suo, F. Ma, Y. Yuan, M. Huai, W.
Zhong, A. Zhang, et al., "Personalized
57
disease prediction using a CNN-based
similarity learning method", Proc. IEEE
Int. Conf. Bioinf. Biomed. (BIBM), pp.
811-816, Nov. 2017.
18. T. Pham, T. Tran, D. Phung and S.
Venkatesh, "DeepCare: A deep dynamic
memory model for predictive
medicine", Proc. Pacific–Asia Conf.
Knowl. Discovery Data Mining,pp.30-
41
18. Kavitha, R , Dr.R.Nedunchelian,
“Domain-Specific Search Engine
Optimization using Healthcare
Ontology and a Neural Network
Backpropagation Approach” , 2017,
ResearchJournal of Biotechnology,
Special Issue 2, Page No:157-16
19. Kavitha, G., Kavitha, R., “An
analysis to improve throughput of high-
power hubs in mobile ad hoc network” ,
2016, Journal of Chemical and
Pharmaceutical Sciences, Vol-9, Issue-2:
361-363
20. Kavitha, G., Kavitha, R., “Dipping
interference to supplement throughput
in MANET” , 2016, Journal of
Chemical and Pharmaceutical Sciences,
Vol-9, Issue-2: 357-360

58
59