Добавил:
Sekretar
kiopkiopkiop18@yandex.ru
t.me/Prokururor I Вовсе не секретарь, но почту проверяю
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз:
Предмет:
Файл:Ординатура / Хирургия / Библиотека им академика М.И. Перельмана / Книга_145_библиотеки_им_акад_М_И_Перельмана
.pdf
Mining Standardized EHR Data Disease Prediction using Machine Learning 153
https://t.me/med1917
depicts the dataset's several entities, the 'Attribute' section includes the numerous
attributes related to each entity, and the 'Value' section provides the attribute's
value. EAV model offers great flexibility for adding, deleting or modifying the
attributes in the database because its logical schema is different from the physical
schema. The different EAV schemas available are as follows [18]:
1. Basic EAV Schema: The physical schema consists of at least three columns
namely entity, attribute and value. The entity column contains the distinct
identifier of the entity. The label or identification of the attribute for whom the
value is to be saved is contained in the 'Attribute' field. Value contains the value
of the attribute. A logical schema is defined using a separate table called a
Metadata table. To store the data which is not sparse, a separate relational table is
used and for other data, an EAV table is used. The transformation of the EAV
table to a relational table is facilitated through the use of a metadata table.
2. Multi Data Types EAV Schema: As the EAV table contains a single column
for storing the value of each attribute, the data type assumed value column is a
string. There are many issues that arise due to the usage of string data type like
when “12” and “2” are compared “12” will have a lower value than “2” in case of
string but numerically “12” is greater than “2”. To deal with such a problem, one
EAV table is made per data type.
3. Hybrid EAV Schema: By establishing numerous data forms under a single
table, it combines the greatest features of the preceding simple and multi-data type
schemas.
Any of the above-defined three EAV schemas can be used depending upon the
requirement. When the EAV model is used, multiple origins and multiple formats
problem can be resolved as EAV is quite flexible to store any kind of physical
schema. This is accomplished via metadata, which contains information on the
logical structure of the data being saved as an EAV model. The summary of
identified challenges is shown in Table 2.
Table 2. Summary of challenges identified and proposed solution.
S. No. Challenge Identified The Proposed Solution for Overcoming
1. Multiple origins EAV model
2. Multiple formats EAV model
1. Replacing missing values with some constants.
3. Missing data
4. Distinguished users EAV model
2. Identifying patterns of occurrence of missing values and replacing missing
values accordingly.
3. Removal of noise through appropriate methods.

154 Disease Prediction using Machine Learning Batra et al.
https://t.me/med1917
(Table 2) co nt.....
S. No. Challenge Identified The Proposed Solution for Overcoming
5. Data granularity EAV model
6. Flexibility EAV model
7. Sparseness EAV model
Issues related to various types of users and different granularity levels can also be
resolved through the EAV model as the attribute list is not fixed and new
attributes can be added easily hence resolving the problem of volatility also.
The entries which contain null values need not be stored when data is stored as the
EAV model because the EAV model does not define the logical schema. So, by
using the EAV model, sparseness can easily be removed.
The EAV model is efficient for storing EHR data but many applications such as
graphing and statistical analysis require data to be present in the legacy relational
row-oriented format. So EAV modelled data needs to be transformed into roworiented data so that DM can be implemented easily on the dataset and this
transformation process is called pivoting [19]. Dinu et al. tested several pivoting
strategies [19]. This comparison helps us to find the best pivoting technique.
Next, to address privacy concerns, EAV includes the implementation of rolebased access control (RBAC), data encryption, anonymization, and auditing. Such
appropriate access controls and security measures at the database level help to
maintain data privacy. This protected privilege refers to limiting access to
sensitive or confidential information only to authorized individuals or roles. This
can be achieved through careful design of access controls, defining appropriate
roles and permissions, and implementing encryption or other security measures.
Thus, it is evident that the EAV model is a useful tool for managing flexible
attributes, addressing privacy concerns, and protecting privileges to provide
additional security measures and access controls in the database system.
RELATED WORK
Managing a huge volume of health data manually is difficult and prone to
mistakes. Furthermore, it is conceivable that concealed information would not be
discovered. Data mining is critical in giving solutions to all of these issues. Chen
et al. [20] have offered a comprehensive understanding of relevant DM methods
as well as the objectives and problems in developing DM strategies. It also gives a
quick rundown of the procedure, as well as the boundaries and benefits of several
DM strategies.

Mining Standardized EHR Data Disease Prediction using Machine Learning 155
https://t.me/med1917
Bellazzi and Zupan [21] present a comprehensive overview of predictive
modeling. This document describes many strategies for predictive DM that are
extensively employed. It also displays all of the objectives that are evaluated
using different categorization approaches. They also gave a quick overview of the
relationship between statistics and predictive DM. Bellazzi and Zupan offered an
outline of predictive DM's contribution to healthcare as well as the procedure
under which it is executed.
In a study [1], Koh and Tan discuss many ways in which health DM might be
beneficial. They also provide a thorough understanding of the difficulties that
might arise when using DM to investigate health records. They give an indication
of the entire DM implementation stage (business comprehension, data
interpretation and anticipation, modelling, assessment, and rollout), as well as the
classification of DM tasks, the advantages of using DM techniques in the health
sector, and the complexities of using DM to assess health records.
The existence of incomplete data values in a database might lead to inaccurate
results when used in an investigation. As a result, a technique to determine the
reason for absent data is required. The research reported [22] tries to detect
patterns (bias) that may be utilized to determine the reason for incomplete
information, i.e. a set of certain attribute-value pairs whose existence corresponds
to a data point in a specific database feature.
Due to the sheer peculiarity of health data, Cios and Moore [16] offer a thorough
assessment of all difficulties that a DM researcher should be aware of before
beginning his investigation.
In a study [18], El-Sappagh et al. describe an efficient framework that can store
connections across distinct categories while also coping with sparseness and
flexibility. EAV can store sparse and flexible data, but it can't deal with
connections between classes. In circumstances when one parameter is an entity of
a different category, the EAV and metadata tables are maintained as entity
relation tables rather than conventional relational tables. The EAV/CR model is
the result of this addition to the EAV model.
In another study [23], Dinu et al. provide a comprehensive overview of the EAV
paradigm. It contains all of the information needed to implement the EAV model
in a solution. The EAV model, which overcomes the constraints of the tabular
data structure, is described in full in this study. EAV modelling can handle all of
the problems that come with tabular data. Each characteristic in an EAV-modelled
system has a matching row that contains entries such as identifier, attribute
identifier, and attribute value.

156 Disease Prediction using Machine Learning Batra et al.
https://t.me/med1917
In another study [19], Dinu et al. compare three pivoting strategies to determine
which is the optimum. By monitoring the effectiveness of three pivoting methods
on different DBMSs, the best approach is determined.
The present research examined the use of DM on EHRs, special aspects of EHRs,
the optimum data structure to record EHRs, the EAV model, and pivoting. The
present investigation analyzes all of these features in order to recognize issues
with DM on EHR applications as well as solutions to these issues.
All the studies that have been conducted so far and reported above have been
conducted separately in various disciplines.
CONCLUSION
Data analysis in medicine particularly for EHRs is a requirement of the present
scenario. The healthcare professional can benefit from the hidden knowledge
collected by data mining techniques. As a result, the present investigation
integrates prior studies on analyzing EHR databases. It establishes that obtaining
meaningful insights from health data may be extremely beneficial to humanity.
Mined information can be analyzed to deliver better treatment and prompt illness
diagnosis, lowering healthcare expenditures. DM has a plethora of advantages like
medical insurance. DM may be used to anticipate future expenditures based on
historical financial information, allowing insurance companies to make more
effective decisions. It further explores how DM can be applied to the EHRs
domain, simultaneously identifying the challenges and proposing solutions for
identified issues.
The foremost challenges in implementing DM techniques in healthcare records
are dealing with sparseness, volatility, privacy, and heterogeneity. All these issues
can be handled by storing the data using the EAV model. Although the use of the
EAV model solves the problem of sparseness, volatility and heterogeneity data
need to be in a relational format so that DM methods can be implemented on it.
So, pivoting techniques are used to convert EAV-modelled data into relationally
modelled data. Once the data is in a uniform format, EAV supports efficient
storage, processing, and controlled access.
REFERENCES
[1] H.C. Koh, and G. Tan, "Data mining applications in healthcare", J. Healthc. Inf. Manag., vol. 19, no.
2, pp. 64-72, 2005.
[PMID: 15869215]
[2] P. Giudici, and Figini Silvia, Applied data mining: Statistical methods for business and industry. 2
Edition. John Wiley & Sons, 2009, p. 264.
[3] S. Batra, and S. Sachdeva, "Organizing standardized electronic healthcare records data for mining",
Health Policy Technol., vol. 5, no. 3, pp. 226-242, 2016.
nd

Mining Standardized EHR Data Disease Prediction using Machine Learning 157
https://t.me/med1917
[http://dx.doi.org/10.1016/j.hlpt.2016.03.006]
[4] F. E. Bekri, and A. Govardhan, "Association of data mining and healthcare domain: Issues and current
state of the art", Global Journal of Computer Science and Technology., 2011.
[5] P. Yadav, M. Steinbach, V. Kumar, and G. Simon, "Mining electronic health records (EHRs) A
survey", ACM Comput. Surv., vol. 50, no. 6, pp. 1-40, 2018.
[http://dx.doi.org/10.1145/3127881]
[6] S. Batra, and S. Sachdeva, "Pre-processing highly sparse and frequently evolving standardized
electronic health records for mining", In: Handbook of Research on Disease Prediction Through Data
Analytics and Machine Learning. IGI Global, 2021, pp. 8-21.
[http://dx.doi.org/10.4018/978-1-7998-2742-9.ch002]
[7] P. Schloeffel, "Electronic Health Record Definition, Scope and Context", ISO. TC. 215., 2002.
[8] Available at: http://www.hl7.org/implement/standards/ansiapproved.cfm
[9] Available at: https://en.wikipedia.org/wiki/CEN/TC_251
[10] Available at: https://www.iso.org/standard/40784.html
[11] Available at: https://www.iso.org/standard/62305.html [Accessed 22 June 2021].
[12] Available at: http://www.openehr.org/
[13] S. Sachdeva, and S. Bhalla, "Semantic interoperability in standardized electronic health record
databases", ACM J. Data Inf. Qual., vol. 3, no. 1, pp. 1-37, 2012.
[http://dx.doi.org/10.1145/2166788.2166789]
[14] A. Tsymbal, and N. Bolshakova, "Guest editorial introduction to the special section on mining
biomedical data", IEEE Trans. Inf. Technol. Biomed., vol. 10, no. 3, pp. 425-428, 2006.
[http://dx.doi.org/10.1109/TITB.2006.876197]
[15] J. Han, and M. Kamber, Data mining concepts and techniques. Morgan Kaufmann Publishers: San
Francisco, CA, 2001, pp. 335-391.
[16] K.J. Cios, and G. William Moore, "Uniqueness of medical data mining", Artif. Intell. Med., vol. 26,
no. 1-2, pp. 1-24, 2002.
[http://dx.doi.org/10.1016/S0933-3657(02)00049-0] [PMID: 12234714]
[17] G. Duftschmid, T. Wrba, and C. Rinner, "Extraction of standardized archetyped data from Electronic
Health Record systems based on the Entity-Attribute-Value Model", Int. J. Med. Inform., vol. 79, no.
8, pp. 585-597, 2010.
[http://dx.doi.org/10.1016/j.ijmedinf.2010.04.007] [PMID: 20537942]
[18] S.H. El-Sappagh, S. El-Masri, A.M. Riad, and M. Elmogy, "Electronic health record data model
optimized for knowledge discovery", International Journal of Computer Science Issues, vol. 9, no. 5,
p. 329, 2012.
[19] V. Dinu, P. Nadkarni, and C. Brandt, "Pivoting approaches for bulk extraction of
Entity–Attribute–Value data", Comput. Methods Programs Biomed., vol. 82, no. 1, pp. 38-43, 2006.
[http://dx.doi.org/10.1016/j.cmpb.2006.02.001] [PMID: 16556470]
[20] H. Chen, S.S. Fuller, C. Friedman, and W. Hersh, "Knowledge management, data mining, and text
mining in medical informatics", In: Medical Informatics. Springer: Boston, MA, 2005, pp. 3-33.
[http://dx.doi.org/10.1007/0-387-25739-X_1]
[21] R. Bellazzi, and B. Zupan, "Predictive data mining in clinical medicine: Current issues and
guidelines", Int. J. Med. Inform., vol. 77, no. 2, pp. 81-97, 2008.
[http://dx.doi.org/10.1016/j.ijmedinf.2006.11.006] [PMID: 17188928]
[22] M.C. Tremblay, K. Dutta, and D. Vandermeer, "Using data mining techniques to discover bias patterns
in missing data", ACM J. Data Inf. Qual., vol. 2, no. 1, pp. 1-19, 2010.
[http://dx.doi.org/10.1145/1805286.1805288]

158 Disease Prediction using Machine Learning Batra et al.
https://t.me/med1917
[23] V. Dinu, and P. Nadkarni, "Guidelines for the effective use of entity–attribute–value modeling for
biomedical databases", Int. J. Med. Inform., vol. 76, no. 11-12, pp. 769-779, 2007.
[http://dx.doi.org/10.1016/j.ijmedinf.2006.09.023] [PMID: 17098467]

Disease Prediction using Machine Learning, 2024, 159-171 159
https://t.me/med1917
CHAPTER 10
Role of Database in Epidemiological Situation
Kanika Soni1, Shelly Sachdeva1 and Shivani Batra
1
National Institute of Technology, Delhi, India
2
KIET Group of Institutions, Delhi-NCR, Ghaziabad, Uttar Pradesh, India
Abstract: In this technological era, the technology of databases is very essential to
many aspects of modern life. To give the prospective medical practitioner, the finest in
class and most recent medical knowledge, it seems mandatory that education in the
health domain be well-integrated with the most recent databases. This is because there
is a growing demand for it and there are benefits from the collaboration of healthrelated issues of the public and database technology. Database technology can help
improve health in several ways, including connecting geographically separated health
providers and patients, collecting data for research studies like drug and vaccine trials,
keeping track of chronic diseases, and guaranteeing that patients follow their prescribed
treatments. In this pandemic situation of COVID-19, which the whole world is
currently suffering, the current paper attempts to emphasize the databases’ role. It
illustrates how the COVID-19 Dataset can be stored, queried, and analyzed, and helps
in providing decision support to various end-users. We have performed descriptive
analysis by executing specific queries on the COVID-19 Dataset. Then, we performed
predictive analysis using two data analysis techniques on the COVID-19 Dataset to
approximate the situation in some major cities of India. Further, we have visualized our
results to get valuable information from our analysis.
2,*
Keywords: Big data, COVID-19, Data analysis, Data visualization, Database
usability, Epidemiological queries, Epidemiological situation.
INTRODUCTION
The technology of databases has advanced significantly over the past few decades,
and we can now perform increasingly more complicated queries on larger data
sets relatively effectively. However, if we analyze the methodology in which the
information is generated, accessed, altered and shared today, we can identify that
a significant portion of the world’s data is still outside the database systems and
the worst part is that we discover a military of the database administrators and
other professionals having technical expertise aiding users to import data into and
extract it out of a database because users are themselves unable to interact directly
*
Corresponding author Shivani Batra: KIET Group of Institutions, Delhi-NCR, Ghaziabad, Uttar Pradesh, India;
E-mail: ms.shivani.batra@gmail.com
Geeta Rani, Vijaypal Singh Dhaka & Pradeep Kumar Tiwari (Eds.)
All rights reserved-© 2024 Bentham Science Publishers

160 Disease Prediction using Machine Learning Soni et al.
https://t.me/med1917
with the database due to a lot of factors as majority of the users are naive and can't
understand the database schema efficiently if it is complex. Users' ability to access
the web directly has been greatly aided by search engines. As a result, users are
now able to get information into and out of the constantly evolving web with
speed and efficiency. Contrarily, the database community created rigid, precisely
defined, and meticulously planned databases under the assumption that the
information would be clear, rigid, and well-structured. As a result, databases are
now challenging to design, challenging to modify, and even more challenging to
query.
Role of Data
Data plays an important role in understanding and managing an epidemiological
situation. The key roles of data in epidemiology are described below:
• Surveillance: Data collection and analysis are crucial for continuously
monitoring the spread of diseases, identifying trends of spreading a disease in a
community, and detecting outbreaks. This data includes lab reports, other
modalities such as X-rays, Ultrasound, CT scan reports, history of
hospitalizations, and mortality rates.
• Risk Assessment: Data helps epidemiologists assess the risk factors associated
with a disease, such as age, gender, geographic location, medical facilities
available, vaccination, population . This information aids in identifying the rate of
disease spread and populations prone to diseases. Based on this analysis, health
experts can guide the public about ways to minimize the disease spread.
• Contact Tracing: Data related to visitors from one place to another along with
their medical history helps the health experts to identify the source of spread and
minimize the spread. For example, the same has been done during COVID-19 to
notify potentially exposed individuals. In case someone was diagnosed with
COVID-19 and met with people in a gathering or personally, then he could notify
on a social platform or personal mode of communication to take precautionary
measures. This helps in minimizing the chances of disease outbreaks.
• Modeling and Predictions: Epidemiological models utilize data to forecast the
future course of an outbreak, estimate disease transmission dynamics, and
evaluate the potential impact of interventions. For example, as a part of another
research we employed the Susceptible, Exposed, Infected, and Recovered (SEIR)
model to simulate the impact of vaccination on disease outbreaks. Such models
assist policymakers in hospital resource management and reduce the burden on
the health industry. Also, such models reduce the mortality rate by informing
people at an early stage and reducing the chances of hospitalization.

Database in Epidemiological Situation Disease Prediction using Machine Learning 161
https://t.me/med1917
The amount of data used for above-stated predictions is context-dependent and
may vary based on the nature of the disease, transmission dynamics, model
employed, desired outcome, etc. For example, deep learning models require a
huge amount of data while machine learning algorithms require less amount of
data for making predictions. But, the more the data, the higher will be the
accuracy and reliability of predictions. The data used for epidemiology prediction
may hinder the privacy of users. Thus, it is essential to strike a balance between
data quantity and privacy considerations.
Privacy-preserving techniques such as de-identification, aggregation, and
anonymization can be applied to protect individual identities while using the data
for analysis and decision-making. Also, there is a need for finding the right
balance between data utility and privacy protection through careful consideration
of legal and ethical frameworks, stakeholder engagement, and a transparent
decision-making process.
Role of the Database
Database technology is essential to many aspects of modern life in this
technological age. By employing online resources to learn about diseases, their
signs, symptoms, preventive actions, and general contact information for
professionals who may help when needed, database technology has improved
human lives. To give the prospective medical practitioner the finest in class and
most recent medical knowledge, it seems quite necessary that medical education
be well endowed with the most recent databases. Additionally, the updated
revision of medical databases might help medical experts to make rapid and
accurate decisions with little risk of inaccuracy in today's fast-paced and tough
online world. The epidemiology database is available to assist epidemiologists,
researchers of public health, health administrators, policy makers, educators and
philanthropists in understanding the distribution of diseases, their indicators and
determinants, statistics of morbidity and mortality, disease trends, their causes,
aggravating factors, and preventative measures. The use of electronic health
records, better systems of laboratory for supporting primary and secondary
prevention, collection of data for research work like vaccine and drug trials,
improving informatics of medical systems with the help of evidence-based,
logical, and affordable medication, and surveilling persistent disease conditions
are just a few of the ways that database technology can offer a roadmap for
improving health.
Epidemiology
Epidemiology is described as the scientific, methodical study, which is datadriven, of the distribution (pattern, frequency), drivers (risk factors, causes), and

162 Disease Prediction using Machine Learning Soni et al.
https://t.me/med1917
occurrences that are associated with health in certain populations (neighborhood,
school, city, state, country, global). Additionally, it is the implementation of this
study to the management of health issues. A broad family of viruses known as
coronaviruses can infect both humans and animals and lead to disease. Various
coronaviruses are known to create illness in humans with several respiratory
infections from the ordinary cold to the most fatal health conditions like Middle
East Respiratory Syndrome (MERS) and severe acute respiratory syndrome
(SARS). The coronavirus that is discovered most recently is responsible for the
coronavirus disease COVID-19 -World health organization (WHO). Around the
world, the number of new cases is rising daily. The World Health Organization
declared COVID-19 as a global epidemiological situation on 11 March 2020.
Our goal in this paper is to provide emphasis how database plays a crucial role in
the Epidemiological Situation. In section 2, we will see the journey of databases.
In section 3 of this paper, we give an overview of the Epidemiological Scenario
and Databases. Section 4 of this paper provides the implementation details. Data
analysis and visualization have been done in section 5 of this paper, future work
needed is explained in section 6 of this paper, and then finally section 7 and
section 8 include the conclusion and references, respectively.
JOURNEY OF DATABASES
One of the biggest problems that humanity has ever had to deal with is data
storage, and it still remains. A solution to the file system's shortcomings,
including redundancy, inconsistency, concurrent access by numerous users, and
many others, has been found in databases. The first generation [1] of databases
emerged around the middle of the 1960s and the beginning of the 1970s in order
to provide a practical and effective method of storing and accessing permanent
data through the creation of general-purpose database management systems.
Based on the relationships between the data being that is stored, databases can be
categorized into 3 groups: Relational Databases [2], NewSQL Databases [3], and
NoSQL Databases [4]. In-depth information about the ongoing advancement of
important database technologies and several related database systems is provided
in this paper [4], which is summarised in Table 1.
Furthermore, Big Data is mostly unreliable, dynamic, loud, and interconnected,
which led to special modeling techniques for querying Big Data. In the end, the
results of the analysis must be interpreted by a decision-maker. This typically
entails reviewing all of the presumptions made and going over the analysis again.
These steps form a Big Data Analysis Pipeline [5], as shown in Fig. (1).
Соседние файлы в папке Библиотека им академика М.И. Перельмана
