Багаторівнева система прийняття рішень для прогнозування та рекомендацій щодо захворювань, пов’язаних із серцем
The precise prediction of health-related issues is a significant challenge in healthcare, with heart-related diseases posing a particularly threatening global health problem. Accurate prediction and recommendation for heart-related diseases are crucial for timely and effective treatment solutions. T...
Збережено в:
| Дата: | 2023 |
|---|---|
| Автори: | , |
| Формат: | Стаття |
| Мова: | Англійська |
| Опубліковано: |
The National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute"
2023
|
| Теми: | |
| Онлайн доступ: | https://journal.iasa.kpi.ua/article/view/277354 |
| Теги: |
Додати тег
Немає тегів, Будьте першим, хто поставить тег для цього запису!
|
| Назва журналу: | System research and information technologies |
| Завантажити файл: |  |
Репозитарії
System research and information technologies| _version_ | 1867334432502841344 |
|---|---|
| author | Sharma, Vedna Samant, Surender Singh |
| author_facet | Sharma, Vedna Samant, Surender Singh |
| author_institution_txt_mv | [
{
"author": "Vedna Sharma",
"institution": "Graphic Era (Deemed to be University), Dehradun"
},
{
"author": "Surender Singh Samant",
"institution": "Graphic Era (Deemed to be University), Dehradun"
}
] |
| author_sort | Sharma, Vedna |
| baseUrl_str | http://journal.iasa.kpi.ua/oai |
| collection | OJS |
| datestamp_date | 2024-02-01T21:03:07Z |
| description | The precise prediction of health-related issues is a significant challenge in healthcare, with heart-related diseases posing a particularly threatening global health problem. Accurate prediction and recommendation for heart-related diseases are crucial for timely and effective treatment solutions. The primary objective of this study is to develop a classification model capable of accurately identifying heart diseases and providing appropriate recommendations for patients. The proposed system utilizes a multilevel-based classification mechanism employing Support Vector Machines. It aims to categorize heart diseases by analyzing patient’s vital parameters. The performance of the proposed model was evaluated by testing it on a dataset containing patient records. The generated recommendations are based on a comprehensive assessment of the severity of clinical features exhibited by patients, including estimating the associated risk of both clinical features and the disease itself. The predictions were evaluated using three metrics: accuracy, specificity, and the receiver operating characteristic curve. The proposed Multilevel Support Vector Machine (MSVM) classification model achieved an accuracy rate of 94.09% in detecting the severity of heart disease. This makes it a valuable tool in the medical field for providing timely diagnosis and treatment recommendations. The proposed model presents a promising approach for accurately predicting heart-related diseases and highlights the potential of soft computing techniques in healthcare. Future research could focus on further enhancing the proposed model’s accuracy and applicability. |
| doi_str_mv | 10.20535/SRIT.2308-8893.2023.4.01 |
| first_indexed | 2025-07-17T10:28:06Z |
| format | Article |
| fulltext |
Vedna Sharma, Surender Singh Samant, 2023
Системні дослідження та інформаційні технології, 2023, № 4 7
TIДC
ПРОБЛЕМИ ПРИЙНЯТТЯ РІШЕНЬ ТА
УПРАВЛІННЯ В ЕКОНОМІЧНИХ, ТЕХНІЧНИХ,
ЕКОЛОГІЧНИХ І СОЦІАЛЬНИХ СИСТЕМАХ
UDC 62-50
DOI: 10.20535/SRIT.2308-8893.2023.4.01
A MULTI-LEVEL DECISION-MAKING FRAMEWORK
FOR HEART-RELATED DISEASE PREDICTION
AND RECOMMENDATION
VEDNA SHARMA, SURENDER SINGH SAMANT
Abstract. The precise prediction of health-related issues is a significant challenge in
healthcare, with heart-related diseases posing a particularly threatening global health
problem. Accurate prediction and recommendation for heart-related diseases are
crucial for timely and effective treatment solutions. The primary objective of this
study is to develop a classification model capable of accurately identifying heart dis-
eases and providing appropriate recommendations for patients. The proposed system
utilizes a multilevel-based classification mechanism employing Support Vector Ma-
chines. It aims to categorize heart diseases by analyzing patient’s vital parameters.
The performance of the proposed model was evaluated by testing it on a dataset con-
taining patient records. The generated recommendations are based on a comprehen-
sive assessment of the severity of clinical features exhibited by patients, including
estimating the associated risk of both clinical features and the disease itself. The
predictions were evaluated using three metrics: accuracy, specificity, and the re-
ceiver operating characteristic curve. The proposed Multilevel Support Vector Ma-
chine (MSVM) classification model achieved an accuracy rate of 94.09% in detect-
ing the severity of heart disease. This makes it a valuable tool in the medical field
for providing timely diagnosis and treatment recommendations. The proposed model
presents a promising approach for accurately predicting heart-related diseases and
highlights the potential of soft computing techniques in healthcare. Future research
could focus on further enhancing the proposed model’s accuracy and applicability.
Keywords: healthcare, heart disease, classification model, learning techniques.
INTRODUCTION
In recent years, a large amount of medical data has become available, represent-
ing the healthcare status of patients. This data includes medical reports, test re-
sults, and lab reports. Data mining plays a significant role in healthcare recom-
mendation systems, as it enables healthcare professionals to extract valuable
insights from the data. These insights can be used to provide more accurate and
personalized recommendations to patients [1], [2]. In the current era, people are
facing major health issues due to inactive lifestyles. The online healthcare system
has proven to be beneficial, especially in scenarios like the COVID-19 pandemic,
and has gained significant attention from researchers. Integrating recommender
systems into healthcare can support doctors, medical professionals, and patients.
These systems assist patients in improving their health conditions and adopting a
Vedna Sharma, Surender Singh Samant
ISSN 1681–6048 System Research & Information Technologies, 2023, № 4 8
healthier lifestyle. Healthcare recommendation systems have evolved with various
learning technologies and big data science, which provide online suggestions to
patients regarding their health issues.
Machine Learning’s application in healthcare is poised to revolutionize the
industry, as it enhances capabilities and reduces expenses. This progress empow-
ers healthcare practitioners, including doctors and personnel, to focus more on
delivering improved patient care. Many researchers have contributed to healthcare
recommendation systems for diagnosing various diseases. Previous studies have
employed machine learning classifiers and deep learning techniques to predict
different diseases using large datasets gathered from various healthcare repositories.
Recommender systems face various challenges, such as reliability, accuracy,
dependability, data loss, and issues related to data integrity and quality. To over-
come these challenges, various classification models have been proposed, includ-
ing the usage of different soft computing techniques like machine learning and
deep learning. The significance of this research is to enrich the healthcare dataset
for better prediction of multidisciplinary diseases. This study proposes an intelli-
gent disease classification mechanism that aims to address various issues in exist-
ing systems by predicting the risks associated with diseases. In this study, a dis-
ease classification model is used to predict the risks related to heart conditions.
The multi-classification methodology works accurately and provides better results
in larger healthcare datasets.
The utilization of the fuzzy technique enables the provision of recommenda-
tions to patients based on the severity score. The organization of the paper is as
follows: Section 2 reviews recent works in the healthcare recommendation field,
Section 3 outlines the research methodology employed in this study, and Section
4 presents the dataset and experimental results. Finally, Section 5 concludes the
paper and discusses future plans.
RELATED WORK
Several studies have been conducted on healthcare recommender systems, target-
ing specific diseases, health-related issues, and recommender contexts. These ex-
isting studies have highlighted the need for a comprehensive overview supported
by a healthcare recommendation system in various recommendation scenarios.
Healthcare recommendation systems offer better personalization, increasing user
understanding of their medical conditions [4]. These systems are also concerned
with providing accurate information, assisting users, and ensuring the security and
privacy of patient information. Traditionally, doctors relied on invasive diagnostic
methods to identify heart disease, involving an assessment of the physical exami-
nation findings, medical history of patients’ and investigation into associated
symptoms [5].
Cardiovascular disease, including coronary artery disease, is a major global
cause of death, particularly among middle-aged and elderly individuals. Tradi-
tional diagnostic methods, such as angiography, are expensive and have notable
side effects. To explore alternative approaches, researchers have extensively stud-
ied data mining techniques and machine learning techniques. In one proposed
work for the accurate diagnosis of coronary heart related disease, the performance
of a neural network improved by approximately 10% through the application of
genetic algorithms to optimize the network’s initial weights [6].
A multi-level decision-making framework for heart-related disease prediction and …
Системні дослідження та інформаційні технології, 2023, № 4 9
Diagnosing and treating heart disease becomes extremely challenging in un-
derdeveloped countries, where there is a lack of necessary medical tools and spe-
cialized professionals [7]. Basic and Deep Neural Networks have shown effi-
ciency in exploratory comparative trials, with the deep neural network
outperforming most other methods [8].
Subiksha et al. [9] designed a framework for a medical care system based on
machine learning. The framework, based on a decentralized network, was de-
signed to link various healthcare databases and services.
Sahoo et al. [10] developed an advanced prediction model. Their study in-
troduces an intelligent health recommendation system (HRS) that leverages big
data analytics to implement an efficient health recommendation engine. The pro-
posed intelligent HRS outperforms existing approaches by achieving a lower
mean absolute error (MAE) value, transforming the healthcare industry into a
more personalized paradigm within a telehealth environment.
Archenna et al. [11] proposed a methodology for generating a healthcare
system. They employed big data analytics in the proposed recommendation sys-
tem, demonstrating how and where to apply big data technologies to construct an
efficient patient recommender system. The study emphasized the need for a sys-
tem capable of handling massive amounts of semi-structured and unstructured
patient information, as well as streaming live information about patients from
various social media activities. By utilizing the appropriate machine learning
(ML) tools and simulations offered by Apache Spark, useful insights can be de-
rived from vast amounts of medical information. The proposed health recommen-
dation system could anticipate a patient’s medical condition by assessing their
lifestyle, general medical variables, cognitive medical conditions, and interper-
sonal networks.
Vanisree K. et al. [12] discuss the significance of an early diagnosis of Con-
genital Heart Disease (CHD) and propose a Decision Support System to improve
accuracy and reduce costs. This system was developed using MATLAB’s GUI
feature and incorporates a Backpropagation Neural Network.
Abugabah et al. [13] designed a medical care analyzer system based on data
mining methodology. In this research work, an optimization-based approach
based on neural network was implemented to achieve efficient results. Clinical
information was retrieved and normalized using a min-max normalization tech-
nique. The patient’s condition was examined and categorized as either healthy or
unhealthy. The supervised learning approach utilized the harmonic optimized
modularity neural network.
Mudaliar et al. [14] developed an application programming interface (API)
that utilizes the machine learning algorithms to analyze a user’s symptoms and
diagnose a specific disease. The framework also recommends suitable drugs for
users afflicted with that disease. The prediction of illness probability takes into
account externally observable symptoms such as temperature, cough, headache,
and other indications experienced by a person. The Naive Bayes algorithm was
employed to diagnose the illness based on these signs.
In a study conducted by Yoo et al. [8], a medical care recommendation sys-
tem based on data mining was developed. The proposed system utilized a peer-to-
peer collection and adaptable judgment response. Handheld sensors were em-
ployed to gather public health records regarding various aspects of an individual’s
Vedna Sharma, Surender Singh Samant
ISSN 1681–6048 System Research & Information Technologies, 2023, № 4 10
life, including nutrition, daily routines, sleeping patterns, lifestyle behaviors, and
occupational stress. Validated index data from the P2P-dataset and personal iden-
tification were also considered. A mobile service-based medical care recommen-
dation cellular modem could be designed to enhance the quality of care for patient
user-based health management, reduce medical costs, and improve service per-
ception of quality in the medicinal industry. Table 1 provides a summary of re-
lated work in healthcare recommendation systems.
Based on the literature review, it can be concluded that selecting important
features based on studies such as [16–18; 28–31] and employing a combination of
classifiers as demonstrated in [16–18; 20; 28–31] can significantly enhance the
predictive capabilities of machine learning algorithms for early-stage detection of
heart disease. However, feature selection presents a challenge due to the exponen-
tial increase in complexity as the number of features in the dataset grows. Evalu-
ating all possible feature subsets becomes computationally intensive and imprac-
tical as the number of features increases [32]. Therefore, alternative strategies are
necessary to address this issue.
From Table 1, it is evident that only a limited number of research studies
have focused on optimizing hyper parameters due to the time-consuming process
of tuning multiple machine-learning models to find the best hyper parameters.
However, effectively optimizing classifier hyper parameters can greatly improve
the accuracy of predicting the risk of coronary heart disease. Healthcare recom-
mendation systems can benefit from the simplicity and accessibility of the multi-
class Support Vector Machine (SVM) model, as it has a limited number of hyper
parameters for tuning. This characteristic simplifies the training and optimization
process. In contrast to more complex machine learning algorithms, the proposed
multi classification-based SVM model’s straightforwardness makes it a conven-
ient choice for implementation in healthcare recommendation systems.
T a b l e 1 . Summary of Related Work
Author
References Proposed Work Outcomes &
Limitations
Scope
for future work
Subiksha,
et al., [9]
(2018)
The deep learning-based health ana-
lyzer system. Performance metrics are
Precision and Recall
High
error rate
Need to enhance the
methodology for better
results in information
retrieval
Sahoo et
al., [10]
(2019)
Deep learning-based recommendation
system for healthcare. Performance
metrics are MAE & RMSE
Inefficient
privacy
results
Need to improve results
by resolving security
features
Archenna
et al., [11]
(2017)
Big data-based Healthcare recom-
mendation system
The performance metric is accuracy
Security and
reliability i
ssues
Need to add security
features
Sharma
et al., [15]
(2017)
Information retrieval approach for
healthcare recommendation system.
The performance metric is accuracy
Limited
features
of diseases
Need to add more
features of diseases for
more efficient results
Shah et al.,
[16] (2020)
Healthcare system based on the deep
learning framework. Performance
metrics are Recall, Precision,
Accuracy & F-score
Less
dataset size
Need to use more
datasets
for efficient results
A multi-level decision-making framework for heart-related disease prediction and …
Системні дослідження та інформаційні технології, 2023, № 4 11
Continued Table 1
Author
References Proposed Work Outcomes &
Limitations
Scope
for future work
Sornalak-
shmi et al.,
[17] (2020)
Healthcare system based on Apriori
algorithm. Performance metrics are
Accuracy. Execution time
Reliability
issues
Need to use different
optimization algorithms
for efficient results
Gebremes-
kel et al.,
[19] (2019)
Dynamic data handling approach.
Performance metrics are MAE &
RMSE
Anomalies
in patient
information
Need to overcome
anomalies in the patient
dataset
Yoo et al.,
[20] (2019)
Recommendation system based
on mining. Performance
metrics are Entropy & Gain
Classification
issues
Implement enhance
feature classification
model
Deng
et al., [21]
(2019)
Collaborative filtering-based
Healthcare system.
The performance metric is recall
Challenging
implementa-
tion
Stage analysis
on classified data.
Hui Yuan
et al., [26]
(2018)
A health recommendation system
based on hybrid technique.
The performance metric is Precision
Offline
collected
information
utilised only
Detect only
a few diseases
Efficient for small
datasets only
Gujar et
al., [27]
(2018)
Machine learning-based health
recommender system.
The performance metric is accuracy
Limited
dataset
To predict and
evaluate the health
of real-time cases
PROPOSED METHODOLOGY
In this study, the proposed methodology presents a system consisting of three dif-
ferent levels pertaining to healthcare entities. These levels, namely data collec-
tion, data execution, and output, aim to facilitate improved decision-making for
doctors and patients. The primary objective of the proposed system is to assist
patients in making timely clinical decisions regarding their medical treatment.
The various stages involved in the suggested methodology are depicted in Fig. 1.
The first step involves pre-processing the patient data to handle outliers and ad-
dress missing data. Subsequently, the feature selection method is applied to the
Dataset
Data labeling
Pre-processing
Feature
Selection
Disease
Classification
Risk Prediction Model
Training
Dataset
Testing
Dataset
Fig. 1. Architecture for heart disease prediction model
Vedna Sharma, Surender Singh Samant
ISSN 1681–6048 System Research & Information Technologies, 2023, № 4 12
datasets during the classification stage, and the appropriate feature sets are cho-
sen. These selected features are then utilized to predict one of the four predeter-
mined classes of heart disease. The predicted information is subsequently com-
bined with the patient’s medical record to offer general medical advice based on
the risk level of the disease.
Data Processing
In this phase, data collected data from different recourses undergoes a preprocess-
ing phase. Soft computing techniques have become an essential part of meaning-
ful analysis and obtaining optimal results. To improve the training model’s per-
formance, we eliminated unnecessary data like missing values, repeating records
and outliers. At the top layer, the collected data is initially cleaned and processed,
to enhance the presentation and quality of the data used for model development.
To identify outliers, missing values, and irrelevant data, we have employed the
numerical cleaner filter technique [28]. The processed data was utilized in the fea-
ture selection process. Once the data labeling process is completed, the data is
subsequently partitioned into two segments: the training dataset and the testing
dataset. The training set is utilized for training the classification model, while the
testing set is employed to assess the model’s performance.
Feature Selection
In the feature selection phase, a subset of highly distinguishing features must be
selected for the diagnosis of diseases. In this process discriminating features are
selected for different available classes [30]. The datasets used in this study com-
prise 20 features, among which only a few are pertinent for decision-making in
the disease classification process. To reduce the feature vector to a more manage-
able sample size, a feature selection technique is employed, consisting of two phases.
In the first phase, an attribute selection technique is utilized to evaluate the
features present in the datasets. This technique helps assess the relevance and im-
portance of each feature.
In the second phase, we have employed a search technique to select the op-
timal set of classification models by systematically exploring different combina-
tions of features. The goal is to identify the most effective combination of features
that yields the best performance for the classification task. The proposed selection
method in this study is information gain-based, wherein the entropy for each class
is evaluated [30]. Features that are selected with higher entropy values are more
informative and have a greater contribution to the decision-making process. For
taking the computing decision selected highest Info Gain is calculated as followed
Info Gain(X) = Info(Y) – InfoA(Y),
X = Investigation feature; Y = Featured Dataset.
In dataset D to classify a feature Info required which is calculated as
Info (I) = n
i ii PP1 2 )(log ,
where n = total number of classes; P = probability; D = dataset.
The selection of features based on their information gain is accomplished
through an information gain-based technique in this study. This technique is ap-
plied using class labels as a basis, followed by a ranker search method. The pur-
A multi-level decision-making framework for heart-related disease prediction and …
Системні дослідження та інформаційні технології, 2023, № 4 13
pose of this technique is to determine the relevance of features in the classifica-
tion task and assign them a ranking accordingly.
Prediction
In the next step selected features are classified for the prediction of diseases after
being mapped onto the training model. The classification of diseases is structured
as a multi-class issue, where patient data is classified into four primary categories,
each representing a distinct disease type.
In order to facilitate the training process, we have employed a classification
algorithm based on multilevel Support Vector Machine (SVM) in this study. This
algorithm functions by utilizing an accurate function and high dimension to sepa-
rate class data using a hyper-plane. SVM is optimized for multiple classes to deal
with real-world issues. In multilevel classification, pair-wise classification of
SVM is used to train the given set of data for each pair of given classes.
Once the model is trained, to evaluate the performance and efficiency of the
proposed approach we have applied the trained model to the testing dataset. The
prediction model results are assessed using three primary metrics, namely accu-
racy, sensitivity, specificity and ROC, to determine its performance.
Risk prediction and recommendation model for heart disease
The main goal of this study is to develop a recommendation system that can pro-
vide accurate recommendations based on the severity of diseases related to heart.
The proposed recommendation model evaluates the patient’s data to predict the
level of risk and determine the severity of the disease. The algorithm for the pro-
posed model is given below:
Input: P = Prediction of Disease, D = Dataset of patients’
Output: Recommendation R: (1, 2, 3, and 4)
(1: No recommendation, 2: Need to normal exercise, 3: Need to visit doctor,
4: Need to get hospitalized and have proper treatment)
1. X = (x1, x2, x3), {X represent critical feature set}
2. (x1: Cholesterol, x2: Blood Pressure, x3: Blood sugar)
3. Y = (Critical, Medium, Normal) {Let Y represent severity range of X}
4. Let W represent the weight of X
5. Kb = X, W, Y (Kb= Knowledge base)
6. for each disease P and info from k, d
7. Calculate Probability, Prob (P) and Prob (P)
8. Prob (P): occurrence of disease, Prob (P): absence of disease
9. R = Prob(P)/Prob(P) {R=Estimate Risk}
10. If Y == Critical then,
11. For R< or > 1 AND for Prob <or > 0.30
12. Indentify S and compute final score; {S=Score}Final Score =
m
i ii WS1 )(
13. End
14. Else if Y = medium then,
15. For R> or <1 AND for prob < or > 0.30 then,
16. FS (Final Score) =
m
i ii WS1 )(
Vedna Sharma, Surender Singh Samant
ISSN 1681–6048 System Research & Information Technologies, 2023, № 4 14
17. End
18. FS (Final Score) = (0 – 5),
19. 0<FS<1.9: R = 1
2< FS <2.9< R = 2
2.9<FS<3.9< R =3
4<FS<5<R = 4
20. End
21. Return R
Recommendation
Scopes of parameters are identified after the prediction of diseases which also
depends upon ranges and risk factor values of severity. The four major general
types of recommendations assigned to patients are:
1. No recommendation.
2. Normal Exercise.
3. Visit to doctor.
4. Need to get hospitalized.
The range of parameters in the feature set for heart disease is x1 = Blood
Pressure x2: Cholesterol, x3: Blood sugar.
The recommendation classes with score ranges and parameters ranges are
given in Table 2 and Table 3.
T a b l e 2 . Class labels and ranges recommendation
Class Labels Scores
Class 1 No recommendation 0–0.25
Class 2 Normal Exercise 0.25–0.50
Class 3 Visit to doctor 0.50–0.75
Class 4 Need to get hospitalized 0.75–1
RESULT AND ANALYSIS
In the following subsection, we provide details about the dataset used and present
the outcomes of the proposed system obtained through experimental evaluation.
Dataset: For our study, we incorporated a heart disease-related dataset. Our
developed system was trained and tested on a heart disease dataset that is openly
accessible in the UCI library at http://archive.ics.uci.edu/ml/datasets/heart+disease.
This dataset consists of approximately 1000 patients’ health records, with health
features described in Table 4.
T a b l e 3 . Ranges of parameters to check heart disease
Ranges
Parameters Weightage
Critical Normal
Blood pressure 0.75 >160 120–160
Cholesterol 0.50 >300 200–300
Blood Sugar 0.25 >125 100–124
A multi-level decision-making framework for heart-related disease prediction and …
Системні дослідження та інформаційні технології, 2023, № 4 15
T a b l e 4 . Parameter for the health status of patients
Blood Pressure
Range of BP (mm Hg) Risk of disease
90/60 (low) High
120/80 (Normal) Fit (No Disease)
140/190 (High) Very High
Cholesterol
Range of Cholesterol (mg/dL) Risk of heart disease
100 to 129 mg/dL Fit (No Disease)
130 to 159 mg/dL Border Line
160 to 189 mg/dL High
190 mg/dL and above Very High
Output for classification phase
Fig. 2 illustrates the ROC curve representing the performance of the predicted
model. The prediction of heart disease achieved an AUC of 0.93 for the MSVM
algorithm. The ROC curve of the Random Forest (RF) model is closer to 1, indi-
cating higher accuracy. The curve showcases the pair of specificity and sensitivity
values for a specific threshold decision at each point.
The performance of the MSVM classifier was evaluated using various per-
formance metrics, primarily accuracy. It was compared to the accuracy of other
existing models, namely KNN (85.1%), Naïve Bayes (89.7%), and neural network
(91.8%). KNN relied on a single parameter, K (number of neighbours), while Naive
Bayes utilized two hyperparameters, α and β, for features classification. The neural
network employed the sigmoid activation function to optimize parameters [33].
Fig. 3 presents a curve that represents the accuracy of the multilevel Support
Vector Machine (MSVM) classification model. The model achieved an accuracy
rate of 94.09%, indicating an optimal solution for improving accuracy compared
to existing models such as KNN, Naïve Bayes, and neural network, as shown in
Figs. 4, 5 and 6 display the specificity and sensitivity rates, respectively. These
Fig. 2. ROC curve for the prediction algorithm
Vedna Sharma, Surender Singh Samant
ISSN 1681–6048 System Research & Information Technologies, 2023, № 4 16
figures provide insights into the model’s performance in terms of correctly identi-
fying true negatives (specificity) and true positives (sensitivity).
Fig. 4. The Accuracy of different algorithms
Fig. 3. Accuracy curve for the prediction algorithm
Fig. 5. Specificity Rate
A multi-level decision-making framework for heart-related disease prediction and …
Системні дослідження та інформаційні технології, 2023, № 4 17
The output of the prediction and recommendation phase
In this section, outcome from the prediction and recommendation phase is pre-
sented. This research work aims to predict health-related issues in patients and
provide recommendations based on the risk and probability of occurrence of dis-
eases. Initially, knowledge is gathered from medical experts and clinical records,
and then the weight of each parameter is applied based on its significance. Based
on the criticality of parameters different ranges are assigned. Parameters falling in
the normal category are ignored.
The following equation is used to evaluate the risk based on the knowledge
base database:
Risk (R) = P (e)/P (e);
P (e) = Probability of disease with abnormality;
P (e) = Probability of disease without abnormality.
The prediction of the diagnosed disease likelihood is based on the input from
the fuzzy system and the classification process and determines the corresponding
risk level. By analyzing the given dataset using this method, the resulting values
indicate whether the patient’s health is at risk or not. The output obtained from the
fuzzy model can be found in Table 5.
T a b l e 5 . Sample table of outputs of recommendation model
Patient id Disease Level Recommendation
51 Normal 0 No recommendation
57 Heart disease 0.25 Normal Exercise
63 Normal 0 No recommendation
75 Critical Heart disease 0.5 Visit to doctor
81 Highly Emergency 1 Need to get hospitalized
CONCLUSION & FUTURE WORK
The study proposes a framework for predicting and recommending treatments for
heart-related diseases using a multilevel decision-making approach. The proposed
Fig. 6. Sensitivity Rate
Vedna Sharma, Surender Singh Samant
ISSN 1681–6048 System Research & Information Technologies, 2023, № 4 18
multilevel classification model, based on the support vector machine, achieves an
accuracy rate of 94.09% and provides crucial health recommendations for the
early prevention of critical diseases to patients upon disease detection. Imple-
menting these recommendations can help reduce the risk of heart disease and en-
sure better health outcomes for patients.
In the proposed work, a dataset of around 1000 patients was used. However,
it should be noted that the proposed model has limitations, as it relies on clinical
validations for any health recommendation decisions.
The proposed model contributes to the medical field by enabling better deci-
sion-making for patient healthcare. The analysis of the results focused on accu-
racy, and we compared the accuracy of our proposed model with three commonly
used algorithms (KNN, Naïve Bayes, and Neural Network). We found that the
MSVM classification algorithm provided the optimal solution with better accuracy.
In future research, the suggested approach will be evaluated using a real-
time dataset. Additionally, the scheme can be expanded by examining the influ-
ence of additional characteristics on the recommendation of heart disease, thereby
enhancing safety during the validation stage. To achieve this, the implementation
of deep learning based technology is necessary. This technology will enable visu-
alization and recording of the learning process, ensuring transparency in the use
of deep learning-based techniques.
REFERENCE
1. Y. Li, and T. Beaubouef, “Data Mining: Concepts, Background, and Methods of Inte-
grating Uncertainty in Data Mining,” CCSC: SC Student E-Journal, 3, pp. 2–7, 2010.
2. Data Mining Process: Models, Process Steps & Challenges Involved (2021). Re-
trieved 7 July 2021. Available: https://www.softwaretestinghelp.com/data-mining-
process/#Steps_In_The_Data_Mining_Process
3. I. Badash et al., “Redefining Health: The Evolution of Health Ideas from Antiquity
to the Era of Value-Based Care,” Cureus, 9(2), e1018, 2017. doi:
10.7759/cureus.1018
4. Robin De Croon, Leen Van Houdt, Nyi Nyi Htun, Gregor Štiglic, Vero Vanden
Abeele, and Katrien Verbert, “Health Recommender Systems: Systematic Review,”
Journal of Medical Internet Research, vol. 23, no. 6, e18035, 2021. doi:
10.2196/18035.
5. Q. An, S. Rahman, J. Zhou, and J.J. Kang, “A Comprehensive Review on Machine
Learning in Healthcare Industry: Classification, Restrictions, Opportunities and
Challenges,” Sensors, 23, 4178, 2023. Available: https://doi.org/10.3390/s23094178
6. Z. Arabasadi, R. Alizadehsani, M. Roshanzamir, H. Moosaei, and A.A. Yarifard,
“Computer aided decision making for heart disease detection using a hybrid neural
network-Genetic algorithm,” Comput. Methods and Programs in Biomed., 141,
pp. 19–26, 2017.
7. H. Yang and J.M. Garibaldi, “A hybrid model for automatic identification of risk
factors for heart disease,” J. Biomed. Inform., 58, pp. 171–182, 2015.
8. Jhonny Pincay et al., “Health Recommender Systems: A State-of-the-Art Review,”
2019 Sixth International Conference on eDemocracy & eGovernment (ICEDEG
2019), pp. 47–55.
9. K. Subiksha, “Improvement in analyzing healthcare systems using deep learning ar-
chitecture,” in 2018 4th International Conference on Computing Communication
and Automation (ICCCA), IEEE, 2018, pp. 1–4.
10. A.K. Sahoo, S. Mallik, C. Pradhan, B.S.P. Mishra, R.K. Barik, and H. Das, “Intelli-
gence-based health recommendation system using big data analytics,” Big data ana-
lytics for intelligent healthcare management, Academic Press, USA, 2019, pp. 227–246.
A multi-level decision-making framework for heart-related disease prediction and …
Системні дослідження та інформаційні технології, 2023, № 4 19
11. J. Archenaa, E.A. Mary Anita, “A health recommender system using big data analytics,”
Journal of Management Science and Business Intelligence, 2(2), pp. 17–24, 2017.
12. K. Vanisree and J. Singaraju, “Decision support system for congenital heart disease
diagnosis based on signs and symptoms using neural networks,” Int. J. Comput.
Appl., 19, pp. 6–12, 2015.
13. A. Abugabah, A. AlZubi, F. Al-Obeidat, A. Alarifi, and A. Alwadain, “Data mining
techniques for analyzing healthcare conditions of urban space-person lung using
meta-heuristic optimized neural networks,” Cluster Computing, vol. 23, no. 3,
pp. 1781–1794, 2020. doi: 10.1007/s10586-020-03127-w.
14. V. Mudaliar, P. Savaridaasan, and S. Garg, “Disease prediction and drug recommen-
dation android application using data mining (virtual doctor),” International Journal
of Recent Technology and Engineering, vol. 8, 2019.
15. D. Sharma, G. Singh Aujla, and R. Bajaj, “Deep neurofuzzy approach for risk and
severity prediction using recommendation systems in connected Healthcare,” Trans-
actions on Emerging Telecommunications Technologies, e4159, 27 October 2020.
16. A. Shah, X. Yan, S. Shah, and G. Mamirkulova, “Mining patient opinion to evaluate
the service quality in healthcare: a deep-learning approach,” Journal of Ambient In-
telligence and Humanized Computing, vol. 11, no. 7, pp. 2925–2942, 2019. doi:
10.1007/s12652-019-01434-8.
17. M. Sornalakshmi et al., “Hybrid method for mining rules based on enhanced Apriori
algorithm with sequential minimal optimization in the healthcare industry,” Neural
Computing and Applications, 2020. doi: 10.1007/s00521-020-04862-2.
18. O.W. Samuel, G.M. Asogbon, A.K. Sangaiah, P. Fang, and G. Li, “An integrated
decision support system based on ANN and Fuzzy_AHP for heart failure risk predic-
tion,” Expert Syst. Appl., 68, pp. 163–172 2017.
19. G.B. Gebremeskel, B. Hailu, and B. Biazen, “Architecture and optimization of data
mining modeling for visualization of knowledge extraction: patient safety care,”
Journal of King Saud University-Computer and Information Sciences, 2019.
20. H. Yoo and K. Chung, “Mining-based life care recommendation using peer-to-peer
dataset and adaptive decision feedback,” Peer-to-Peer Networking and Applications,
vol. 11, no. 6, pp. 1309–1320, 2017. doi: 10.1007/s12083-017-0620-2.
21. X. Deng and F. Huangfu, “Collaborative Variational Deep Learning for Healthcare
Recommendation,” IEEE Access, vol. 7, pp. 55679–55688, 2019. doi:
10.1109/access.2019.2913468.
22. A. Sahoo, C. Pradhan, R. Barik, and H. Dubey, “DeepReco: Deep Learning Based
Health Recommender System Using Collaborative Filtering,” Computation, vol. 7,
no. 2, p. 25, 2019. doi: 10.3390/computation7020025.
23. S. Rathore, M. Habes, M.A. Iftikhar, A. Shacklett, and C. Davatzikos, “A review of
neuroimaging-based classification studies and associated feature extraction methods
for Alzheimer’s disease and its prodromal stages,” Neuroimage, 155, pp. 530–548,
2017. doi: 10.1016/j.neuroimage.2017.03.057.
24. Carolyn Petersen et al., “Recommendations for the safe, effective use of adaptive
CDS in the US healthcare system: an AMIA position paper,” J. Am. Medical Infor-
matics Assoc., 28(4), pp. 677–684, 2021.
25. Khushboo Thaker, “KA-Recsys: Patient-Focused Knowledge Appropriate Health
Recommender System,” SIGIR 2022: Proceedings of the 45th International ACM
SIGIR Conference on Research and Development in Information Retrieval. New
York, USA: Association for Computing Machinery, 2022.
26. Hui Yuan and Weiwei Deng, “Doctor recommendation on healthcare consultation
platforms: an integrated framework of knowledge graph and deep learning,” Internet
Res., 32(2), pp. 454–476, 2022.
27. D. Gujar, R. Biyani, T. Bramhane, S. Bhosale, and T.P. Vaidya, “Disease prediction
and doctor recommendation system,” International Research Journal of Engineering
and Technology (IRJET), 5, pp. 3207–3209, 2018.
Vedna Sharma, Surender Singh Samant
ISSN 1681–6048 System Research & Information Technologies, 2023, № 4 20
28. S.B. Patil and Y. Kumaraswamy, “Intelligent and effective heart attack prediction
system using data mining and artificial neural network,” Eur. J. Sci. Res., 31,
pp. 642–656, 2009.
29. Mark Hall, Eibe Frank, Geoffrey Holmes, Bernhard Pfahringer, Peter Reutemann,
and Ian Witten, “The WEKA data mining software: An update,” SIGKDD Explor.
Newsl., 11, pp. 10–18, 2008.
30. A. Mustaqeem, S.M. Anwar, A.R. Khan, and M. Majid, “A statistical analysis based
recommender model for heart disease patients,” Int. J. Med. Inform., 108, pp. 134–145,
2017. doi: 10.1016/j.ijmedinf.2017.10.008.
31. Pallavi Chavan, Brian Thoms, and Jason T. Isaacs, “A Recommender System for
Healthy Food Choices: Building a Hybrid Model for Recipe Recommendations
using Big Data Sets,” HICSS, pp. 1–10, 2021.
32. Megha Rathi and Vikas Pareek, “Mobile Based Healthcare Tool an Integrated Dis-
ease Prediction & Recommendation System,” International Journal of Knowledge
and Systems Science, 10(1), pp. 38–62, 2019. doi: 10.4018/IJKSS.2019010103.
33. N. Priyanka and Pushpa RaviKumar, “Usage of data mining techniques in predicting
the heart diseases — Naive Bayes & decision tree,” 2017 International Conference
on Circuit, Power and Computing Technologies (ICCPCT). doi:
10.1109/ICCPCT.2017.8074215.
Received 24.04.2023
INFORMATION ON THE ARTICLE
Vedna Sharma, ORCID: 0000-0003-3001-0931, Graphic Era (Deemed to be University)
Dehradun, India, e-mail: sharma.vedna@gmail.com
Surender Singh Samant, ORCID: 0000-0001-8619-3779, Graphic Era (Deemed to be Uni-
versity) Dehradun, India, e-mail: surender.samant@gmail.com
БАГАТОРІВНЕВА СИСТЕМА ПРИЙНЯТТЯ РІШЕНЬ ДЛЯ
ПРОГНОЗУВАННЯ ТА РЕКОМЕНДАЦІЙ ЩОДО ЗАХВОРЮВАНЬ,
ПОВ’ЯЗАНИХ ІЗ СЕРЦЕМ / Ведна Шарма, Сурендер Сінгх Самант
Анотація. Точне прогнозування проблем, пов’язаних зі здоров’ям, є серйоз-
ною проблемою в галузі охорони здоров’я, причому серцево-судинні захворю-
вання становлять особливо загрозу в глобальній проблемі охорони здоров’я.
Точне прогнозування та рекомендації щодо серцево-судинних захворювань
мають вирішальне значення для надання своєчасного та ефективного лікуван-
ня. Основною метою цього дослідження є розроблення моделі класифікації,
здатної точно ідентифікувати захворювання серця та надати відповідні реко-
мендації для пацієнтів. Запропоновано систему, яка застосовує багаторівневий
механізм класифікації з використанням опорних векторних машин. Він спря-
мований на класифікацію захворювань серця шляхом аналізу життєво важли-
вих параметрів пацієнта. Ефективність запропонованої моделі оцінено шляхом
її тестуванням на наборі даних, який містить записи пацієнтів. Сформовані ре-
комендації ґрунтуються на всебічному оцінюванні тяжкості клінічних проявів,
які демонструють пацієнти, включно з пов’язаним ризиком як клінічних ознак,
так і самого захворювання. Прогнози оцінено за трьома показниками: точність,
специфічність і крива робочих характеристик приймача. Запропонована мо-
дель класифікації Multilevel Support Vector Machine (MSVM) досягла рівня то-
чності 94,09% у виявленні тяжкості серцевих захворювань, що робить її цін-
ним інструментом у галузі медицини для надання своєчасної діагностики та
рекомендацій щодо лікування. Запропонована модель дає багатообіцяльний
підхід для точного прогнозування захворювань, пов’язаних із серцем, і засвід-
чує потенціал методів програмного обчислення в сфері охорони здоров’я. По-
дальші дослідження можна зосередити на удосконаленні підвищення точності
та застосовності запропонованої моделі.
Ключові слова: охорона здоров’я, захворювання серця, модель класифікації,
методи навчання.
|
| id | journaliasakpiua-article-277354 |
| institution | System research and information technologies |
| keywords_txt_mv | keywords |
| language | English |
| last_indexed | 2025-07-17T10:28:06Z |
| publishDate | 2023 |
| publisher | The National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute" |
| record_format | ojs |
| resource_txt_mv | journaliasakpiua/31/2209573b5d7c2b9db666c6893bb32131.pdf |
| spelling | journaliasakpiua-article-2773542024-02-01T21:03:07Z A multi-level decision-making framework for heart-related disease prediction and recommendation Багаторівнева система прийняття рішень для прогнозування та рекомендацій щодо захворювань, пов’язаних із серцем Sharma, Vedna Samant, Surender Singh healthcare heart disease classification model learning techniques охорона здоров’я захворювання серця модель класифікації методи навчання The precise prediction of health-related issues is a significant challenge in healthcare, with heart-related diseases posing a particularly threatening global health problem. Accurate prediction and recommendation for heart-related diseases are crucial for timely and effective treatment solutions. The primary objective of this study is to develop a classification model capable of accurately identifying heart diseases and providing appropriate recommendations for patients. The proposed system utilizes a multilevel-based classification mechanism employing Support Vector Machines. It aims to categorize heart diseases by analyzing patient’s vital parameters. The performance of the proposed model was evaluated by testing it on a dataset containing patient records. The generated recommendations are based on a comprehensive assessment of the severity of clinical features exhibited by patients, including estimating the associated risk of both clinical features and the disease itself. The predictions were evaluated using three metrics: accuracy, specificity, and the receiver operating characteristic curve. The proposed Multilevel Support Vector Machine (MSVM) classification model achieved an accuracy rate of 94.09% in detecting the severity of heart disease. This makes it a valuable tool in the medical field for providing timely diagnosis and treatment recommendations. The proposed model presents a promising approach for accurately predicting heart-related diseases and highlights the potential of soft computing techniques in healthcare. Future research could focus on further enhancing the proposed model’s accuracy and applicability. Точне прогнозування проблем, пов’язаних зі здоров’ям, є серйозною проблемою в галузі охорони здоров’я, причому серцево-судинні захворювання становлять особливо загрозу в глобальній проблемі охорони здоров’я. Точне прогнозування та рекомендації щодо серцево-судинних захворювань мають вирішальне значення для надання своєчасного та ефективного лікування. Основною метою цього дослідження є розроблення моделі класифікації, здатної точно ідентифікувати захворювання серця та надати відповідні рекомендації для пацієнтів. Запропоновано систему, яка застосовує багаторівневий механізм класифікації з використанням опорних векторних машин. Він спрямований на класифікацію захворювань серця шляхом аналізу життєво важливих параметрів пацієнта. Ефективність запропонованої моделі оцінено шляхом її тестуванням на наборі даних, який містить записи пацієнтів. Сформовані рекомендації ґрунтуються на всебічному оцінюванні тяжкості клінічних проявів, які демонструють пацієнти, включно з пов’язаним ризиком як клінічних ознак, так і самого захворювання. Прогнози оцінено за трьома показниками: точність, специфічність і крива робочих характеристик приймача. Запропонована модель класифікації Multilevel Support Vector Machine (MSVM) досягла рівня точності 94,09% у виявленні тяжкості серцевих захворювань, що робить її цінним інструментом у галузі медицини для надання своєчасної діагностики та рекомендацій щодо лікування. Запропонована модель дає багатообіцяльний підхід для точного прогнозування захворювань, пов’язаних із серцем, і засвідчує потенціал методів програмного обчислення в сфері охорони здоров’я. Подальші дослідження можна зосередити на удосконаленні підвищення точності та застосовності запропонованої моделі. The National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute" 2023-12-26 Article Article application/pdf https://journal.iasa.kpi.ua/article/view/277354 10.20535/SRIT.2308-8893.2023.4.01 System research and information technologies; No. 4 (2023); 7-20 Системные исследования и информационные технологии; № 4 (2023); 7-20 Системні дослідження та інформаційні технології; № 4 (2023); 7-20 2308-8893 1681-6048 en https://journal.iasa.kpi.ua/article/view/277354/290115 |
| spellingShingle | охорона здоров’я захворювання серця модель класифікації методи навчання Sharma, Vedna Samant, Surender Singh Багаторівнева система прийняття рішень для прогнозування та рекомендацій щодо захворювань, пов’язаних із серцем |
| title | Багаторівнева система прийняття рішень для прогнозування та рекомендацій щодо захворювань, пов’язаних із серцем |
| title_alt | A multi-level decision-making framework for heart-related disease prediction and recommendation |
| title_full | Багаторівнева система прийняття рішень для прогнозування та рекомендацій щодо захворювань, пов’язаних із серцем |
| title_fullStr | Багаторівнева система прийняття рішень для прогнозування та рекомендацій щодо захворювань, пов’язаних із серцем |
| title_full_unstemmed | Багаторівнева система прийняття рішень для прогнозування та рекомендацій щодо захворювань, пов’язаних із серцем |
| title_short | Багаторівнева система прийняття рішень для прогнозування та рекомендацій щодо захворювань, пов’язаних із серцем |
| title_sort | багаторівнева система прийняття рішень для прогнозування та рекомендацій щодо захворювань, пов’язаних із серцем |
| topic | охорона здоров’я захворювання серця модель класифікації методи навчання |
| topic_facet | healthcare heart disease classification model learning techniques охорона здоров’я захворювання серця модель класифікації методи навчання |
| url | https://journal.iasa.kpi.ua/article/view/277354 |
| work_keys_str_mv | AT sharmavedna amultileveldecisionmakingframeworkforheartrelateddiseasepredictionandrecommendation AT samantsurendersingh amultileveldecisionmakingframeworkforheartrelateddiseasepredictionandrecommendation AT sharmavedna bagatorívnevasistemaprijnâttâríšenʹdlâprognozuvannâtarekomendacíjŝodozahvorûvanʹpovâzanihízsercem AT samantsurendersingh bagatorívnevasistemaprijnâttâríšenʹdlâprognozuvannâtarekomendacíjŝodozahvorûvanʹpovâzanihízsercem AT sharmavedna multileveldecisionmakingframeworkforheartrelateddiseasepredictionandrecommendation AT samantsurendersingh multileveldecisionmakingframeworkforheartrelateddiseasepredictionandrecommendation |