Добавил:
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
Скачиваний:
0
Добавлен:
31.08.2026
Размер:
36 Мб
Скачать
Database in Epidemiological Situation Disease Prediction using Machine Learning 163
https://t.me/med1917
Fig. (1). Big Data Analysis Pipeline.
Table 1. Timeline of major database technologies.
Year
1960
Network Databases
1970
1980
Early
2000s
Late
2000s
NoSQL Databases
NewSQL Databases
Database
Technology
Hierarchical and
Relational
Databases
Object-Oriented
Databases
Pros Cons
High performance and
throughput.
By separating the logical
data representation and
physical data
representation, it offers
a significant level of
data independence.
The database and
application programme
are seamlessly
connected.
High availability
and scalability.
Handle the stringent
scalability and reliability
needs of contemporary
OLTP applications.
Application programmes are difficult to create and adapt,
even for straightforward queries,
as a result of the lack of or
restricted data independence.
Mismatch of impedance is
encountered by software
developers, with the growing use
of object-oriented programming
languages for application
programmes.
Due to the significant
expenditures made in relational
databases, it failed to gain
dominance.
Understanding the existing
NoSQL Stores and their various
unique functionalities in-depth is
necessary for the 'Migration
from SQL to NoSQL'.
Doesn’t support full access to
the traditional SQL tools.
Database
Systems
IMS, IDS
Oracle, DB2,
MySQL,
PostgreSQL
Matisse,
ObjectStore,
Versant
Cassandra,
MongoDB, Riak
H-Store,
NuoDB
Almost every area of our contemporary civilization is now impacted by “Big Data,” including trade, governance, health, and research in almost all of the domains, including life sciences, industry, the natural sciences, and the arts and humanities. Big Data is able to revolutionise not only just research but also much more than that. The most thorough enhancement and adoption of Big Data technologies, led by enterprises centered on the Web like LinkedIn, Facebook, Microsoft, Twitter, Quantcast, and Yahoo, can be attributed to the work of Google on Google File System and MapReduce, as well as upcoming open-source work on systems such as Hadoop. Big Data's benefits for healthcare have been argued for with conviction. But many myths surround the term “Big Data,” as highlighted in a study [5]. Data collected from multiple sources need to be filtered and compressed for several reasons. Most data sources are notoriously unreliable, and
164 Disease Prediction using Machine Learning Soni et al.
https://t.me/med1917
the data collected is usually not in a format ready for analysis. Effective large­scale analysis often requires a collection of heterogeneous data from multiple sources. To resolve these heterogeneities, data analysts need a collection of tools for the transformation and integration of the data.
Recent demands for Big Data storage and querying have exposed a number of limitations of conventional relational database systems. As a result, a brand-new category of complementary non-relational data store known as NoSQL has emerged. This category is further divided [4] into Key-Value Stores, Column­Oriented, Document-Oriented, and Graph-Oriented. Table 2 provides a comparison of each of these NoSQL stores.
Table 2. Various NoSQL Stores.
Type Examples Best Fit For
Applications that only require a single key for
Key-Value Redis, Riak
Column-Oriented
Document-Oriented
Graph-Oriented
Cassandra,
MongoDB,
InfiniteGraph
HBase
CouchDB
Neo4j,
accessing data, which include user profiles and
configurations, online shopping carts, and web
Column-oriented stores are highly appropriate
for applications of analytics due to their high
flexibility and scalability as well as their
support for MapReduce operations (for the
parallel processing of huge consolidated
datasets). Web analytics is an example of
application that instruments pages to track the
activities of their visitors.
Applications like blogging platforms and
Content Management Systems (CMS) whose
data could be effortlessly expressed in
document format. For example, the document
format can be used to easily represent a blog
post that contains multiple (nested) attributes -
like comments, photographs, videos and tags.
Sites of social networking involve individuals’
interpersonal contact and social interaction.
session data.
Real-Time
Application
Amazon Dynamo
Facebook’s Inbox
Search
McGraw-Hill
Education’s Self-
Adapting and
Interactive
Learning Portal
Facebook’s Social
Network
Looking back at database history reveals a pattern wherein an increasing variety of data are being processed and stored in databases. This necessitates creating a multi-model database system [7] with the capacity to manage multiple types of data at once. Recent NoSQL database trends point to a shift toward multi-model databases. A thorough taxonomy of several multi-model databases is provided in [7]. Table 3, displays a timeline for the support of the multi-models.
Database in Epidemiological Situation Disease Prediction using Machine Learning 165
https://t.me/med1917
Table 3. Timeline for the support of multi-model databases.
DBMS Original Model Year for support of multi-model database
SQL Server Relational 2000
PostgreSQL Relational 2006
IBM DB2 Relational 2007
CouchDB
Oracle MySQL Relational 2011
Riak Key-Value 2014
Cassandra Column-Oriented 2015
MongoDB Document-Oriented 2016
Cosmos DB Document-Oriented 2017
Document-Oriented
2010
Furthermore, Pandemic data, which is mostly unreliable, dynamic, loud, and interconnected, need querying flexibility to query this huge volume of data efficiently. Ultimately, a global Pandemic situation demands cost-efficient systems which are highly scalable to provide a smooth multiple-user experience. In this paper, we have performed the following steps on the COVID-19 Dataset [8], as shown in Fig. (2). The data storage adopted and the corresponding implementation done by authors in the current research are presented in the next section.
Considering the tremendous development in database technology and its benefits in almost all sectors, whether finance, healthcare, or business, there is a requirement to understand and evaluate the role of databases in such a pandemic situation of COVID-19, where the whole world is currently suffering.
EPIDEMIOLOGICAL SCENARIO AND DATABASES
In today's fast-paced, tough, and online world, the most recent updates to databases of the medical domain can also help experts to make decisions quickly and accurately with a low danger of inaccuracy. The readily accessible epidemiology database would be helpful to epidemiologists, public health experts, health managers, health policymakers, rehabilitation specialists, and health philanthropists in understanding the distribution of diseases, their determinants and indicators, statistics of morbidity and mortality, trends of diseases, and their causes, precipitating 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,
166 Disease Prediction using Machine Learning Soni et al.
https://t.me/med1917
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. Usually, Pandemic data is collected from multiple sources, which need to be filtered and compressed for several reasons. Most data sources are notoriously unreliable, and the data collected is usually not in a format ready for analysis. Collecting heterogeneous data from a variety of sources is frequently necessary for large-scale analysis to be effective. Data analysts require efficient data stores supporting a rich data structure to resolve these heterogeneities.
IMPLEMENTATION DETAILS
As depicted in Fig. (2), as the first step we have stored our dataset in MongoDB. Then, for the second step, we used Python Language to execute all the epidemiological queries. After reading the data from the CSV files of our dataset, we have applied pre-processing to our dataset. Pre-processing or Data Wrangling is an essential step because it is the only way to make raw data usable. Good data wrangling will be able to interpret, clean, and transform data into valuable insights. Pre-processing our dataset included various steps such as we have combined string formed Date and Time to single DateTime, then we drop Sno, Date, Time, replaced ‘-’ in Value columns by NaN, and then replacing ‘Nagaland#’ with ‘Nagaland’ which is a possibly typographical error in the Dataset, we have converted date-time format of the Dataset, and replaced common strings in nationality, e.g., India, Indian, INDIAN are all the same, etc. For the third step, we performed descriptive as well as predictive data analysis. As a final step, we have visualized our results to obtain valuable insights from the epidemiological dataset of COVID-19.
Fig. (2). Steps performed on the COVID-19 Dataset.
Dataset Description
We used a dataset that contained daily information from India's states and union territories. The Ministry of Health & Family Welfare, Government of India [9], is where the state-level data comes from, and [10] is where the individual-level data
comes from. Our dataset contains 5 CSV files, each containing data regarding the following:
1. COVID-19 cases at a daily level.
2. Individual-level details.
Database in Epidemiological Situation Disease Prediction using Machine Learning 167
https://t.me/med1917
3. The population at the state level.
4. The average daily COVID-19 test count.
5. The number of hospital beds in each state.
Query Scenarios
The following Table 4 shows the queries executed by us on the COVID-19 dataset and their corresponding MongoDB queries.
Table 4. Queries Executed on the COVID-19 Dataset.
Query MongoDB Query
Find the total number of cases for each state from Feb 2020 to April 2020. collection[].find().count()
Find confirmed cases, cured cases, and deaths in the top 5 states with the
most number of total cases.
Calculate the mean value of the count of cured patients and the required
number of beds.
Calculate the log value of the required number of beds and the average
number of cured patients.
collection[].aggregate().values()
collection[].aggregate().values()
collection[].aggregate().values()
After the successful implementation of data storage and query scenario identification, the authors performed data analysis and visualization as depicted in Fig. (2).
DATA ANALYSIS AND VISUALIZATION
Data analysis [11] is a means to obtain valuable information through processing data by traditional ideas, technologies, and techniques in statistics, mathematics, or logic, while these theories, processes, and tools that allow for a thorough comprehension of data and the discovery of useful insight within it are known as Data analytics [11]. Data analytics is further divided into various types like descriptive analytics, predictive analytics, prescriptive analytics, explicit analytics, implicit analytics, and deep analytics.
1. Descriptive Analytics: Data analytics uses statistics for explaining the data that is used for generating information or for other several beneficial objectives.
2. Predictive Analytics: Data analytics that, generally using advanced analytics, predicts unknowable future events and reveals the causes behind them.
168 Disease Prediction using Machine Learning Soni et al.
https://t.me/med1917
3. Prescriptive Analytics: Analytics of data, which improves indications and suggests actions for wise decision-making.
4. Explicit Analytics: By means of reporting, descriptive analysis, notifying, and prediction, it focuses on descriptive analytics in general.
5. Implicit Analytics: It focuses on deep analytics, generally using optimization, predictive modelling, prescriptive analytics, and the dissemination of actionable knowledge.
6. Deep Analytics: Data analytics that can gain a thorough grasp of the reasons and circumstances around events, whether they have already occurred, are currently occurring, or will do so in the future, cannot be answered by descriptive analytics.
In this section, we show our descriptive analytics results used to gain information from the epidemiological data. Fig. (3) shows the plot for the number of cases for each state from Feb 2020 to April 2020. Total confirmed cases, cured cases, and deaths in the top 5 states with the most number of total cases are shown in Fig. (4). Fig. (5) shows the results for the average number of cured patients and the number of required beds. Further, Fig. (6) shows the logarithmic fit for the count of cured patients and the required count of beds. Then, we applied two data analysis techniques to present our predictive analytics results.
Fig. (3). Count of Cases for each state from Feb 2020 till April 2020.
Database in Epidemiological Situation Disease Prediction using Machine Learning 169
https://t.me/med1917
Fig. (4). Confirmed, Cured, Deaths for top 5 states.
Fig. (5). Average count of cured patients and required count of beds.
Fig. (6). Logarithmic fit for the count of cured patients and required count of beds.
170 Disease Prediction using Machine Learning Soni et al.
https://t.me/med1917
FUTURE WORK
Around 25 years ago [12] database usability started receiving attention and it has gained more momentum lately [13]. Researchers need to focus on database usability by providing effective and innovative query interface designs which include a keyword, visual, natural language interfaces, and spreadsheet interfaces where the non-technical users (skilled/semi-skilled like nurses, doctors etc.) can query the database using an easy-to-use query language interface. Whether a search engine could simply sit on top of the database systems and is allowed to manage the interface with users by the database systems is one of the obvious questions to raise. We contend that the response is “No”. Although the interface of a search engine is effective for the Web, it does not completely solve the issues related to usability that database systems are facing. This is because of a number of characteristics that result from the expectations of the users for database interaction, which are fundamentally varied from the anticipations for using the Web. For example, users expect to be able to query the database in a more sophisticated manner and want to express the query semantics which are more complex during the interaction with databases because they are aware that databases are not only just the collections of documents but also they contain structures within them which are available for answering their queries. Additionally, more accurate and thorough results are anticipated from the database searches by the users. To overcome all these challenges, we believe that there is a need to provide interactive browsing and navigation query interface [14] which can be on top of the NoSQL databases.
CONCLUSION
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 need for and benefit from the integration of issues related to public health and database technology. In today's fast-paced, tough, and online world, the most recent updates to databases of the medical domain can also help experts to make decisions quickly and accurately with a low danger of inaccuracy. 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
Database in Epidemiological Situation Disease Prediction using Machine Learning 171
https://t.me/med1917
medication, and surveilling persistent disease conditions are just a few of the ways that database technology can offer a roadmap for improving health.
This paper attempted to illustrate the database's critical function in the epidemiological scenario. We used the COVID-19 dataset to conduct a descriptive analysis by running particular queries over it. Then, in order to simulate the scenario in a few key Indian cities, we ran a predictive analysis on the COVID-19 Dataset using two data analysis methodologies. Additionally, in order to extract useful information from our investigation, our results are visualized.
REFERENCES
[1] D.C. Tsichritzis, and F.H. Lochovsky, "Hierarchical data-base management: A survey", ACM Comput.
Surv., vol. 8, no. 1, pp. 105-123, 1976.
[http://dx.doi.org/10.1145/356662.356667] [2] E.F. Codd, A relational model of data for large shared data banks.Software pioneers. Springer: Berlin,
Heidelberg, 2002, pp. 263-294.
[http://dx.doi.org/10.1007/978-3-642-59412-0_16] [3] A. Pavlo, and M. Aslett, "What’s really new with NewSQL?", SIGMOD Rec., vol. 45, no. 2, pp. 45-
55, 2016.
[http://dx.doi.org/10.1145/3003665.3003674] [4] A. Davoudian, L. Chen, and M. Liu, "A survey on NoSQL stores", ACM Comput. Surv., vol. 51, no. 2,
pp. 1-43, 2019.
[http://dx.doi.org/10.1145/3158661] [5] H.V. Jagadish, "Big data and science: Myths and reality", Big Data Research, vol. 2, no. 2, pp. 49-52,
2015.
[http://dx.doi.org/10.1016/j.bdr.2015.01.005] [6] H.V. Jagadish, J. Gehrke, A. Labrinidis, Y. Papakonstantinou, J.M. Patel, R. Ramakrishnan, and C.
Shahabi, "Big data and its technical challenges", Commun. ACM, vol. 57, no. 7, pp. 86-94, 2014.
[http://dx.doi.org/10.1145/2611567] [7] J. Lu, and I. Holubová, "Multi-model Databases", ACM Comput. Surv., vol. 52, no. 3, pp. 1-38, 2020.
[http://dx.doi.org/10.1145/3323214] [8] Available from: https://www.kaggle.com/sudalairajkumar/covid19-in-india [9] Available from: https://www.mohfw.gov.in/ [10] Available from: https://www.covid19india.org/ [11] L. Cao, "Data Science", ACM Comput. Surv., vol. 50, no. 3, pp. 1-42, 2018.
[http://dx.doi.org/10.1145/3076253] [12] C.J. Date, "Database usability", SIGMOD Rec., vol. 13, no. 4, pp. 1-1, 1983.
[http://dx.doi.org/10.1145/971695.582194] [13] H.V. Jagadish, A. Chapman, A. Elkiss, M. Jayapandian, Y. Li, A. Nandi, and C. Yu, "Making database
systems usable", Proceedings of the 2007 ACM SIGMOD international conference on Management of
data, 2007pp. 13-24
[http://dx.doi.org/10.1145/1247480.1247483] [14] M. Kahng, S.B. Navathe, J.T. Stasko, and D.H. Chau, Interactive browsing and navigation in
relational databases. Georgia Institute of Technology: Atlanta, GA, USA, 1603, p. 02371.
172 Disease Prediction using Machine Learning, 2024, 172-177
https://t.me/med1917
SUBJECT INDEX
A
Adaptive 35, 42, 62, 67 neuro-fuzzy inference system 35, 42 threshold technique 67 thresholding technique 62 Algorithm(s) 22, 24, 25, 38, 40, 42, 63, 66, 67,
81, 82, 83, 84, 85, 105, 106, 108, 123,
124, 125, 126, 134 clustering 63 complimentary 81 dataset method findings 123, 124, 125, 126 translation 134 Alzheimer’s disease 67, 80, 89, 90 Angiography 113, 114 green 113 Applications 48, 52, 57, 105, 115, 116, 117,
131, 134, 135, 139, 143, 146, 147, 164 biomedical 52 dermatology 57 Applied data augmentation techniques 7 Artificial 5, 31, 40, 41, 64, 71, 73, 83, 106 intelligence-based techniques 31 neural networks (ANN) 5, 40, 41, 64, 71,
73, 83, 106 Assessment, therapeutic efficacy 146 Asthma 92, 93, 94, 95, 96, 97, 100 attributes 94 comorbidities 94 disease outcome 100 rural 96 urban 97 Autonomic nervous system dysfunction
(ANSD) 39
B
Bayesian networks 72 BCI 131, 132, 143 system processing 143 technology 131, 132, 143 Biomedical 35, 42, 53, 74
Geeta Rani, Vijaypal Singh Dhaka & Pradeep Kumar Tiwari (Eds.)
All rights reserved-© 2024 Bentham Science Publishers
image processing 53 imaging analysis 74 voice measurements 35, 42 Blinking electrooculogram 143 Blogging platforms 164 Blood 56, 119 pressure, systolic 56 vessels, abnormal 119 Blot haemorrhages 118 Bold contrast method 137 Bone(s) 17, 18, 19, 23, 24 deformities 17, 18 finished femoral 23 fractured 19 fracture 18 imaging techniques 17 mineral density (BMD) 18 Brain 52, 53, 54, 67, 131, 132, 133, 134, 135,
138, 139, 142, 143
activity 131, 138
-computer interfaces (BCI) 132, 133, 134, 138, 139, 142, 143
disorders 52
-generated oscillations 135
imaging 67, 131 injury 53, 54 Brain signals 132, 133, 135, 142, 143 distorted 132 non-stationary electrophysiological 142 Brain tumor 62, 67 MRI analysis 67
C
CAD 57, 71, 72, 73, 106
-based machine-learning technique 73
software tools 73 systems 57, 71, 73, 106 techniques 72 CAD tools 55, 56, 71 in healthcare 71 Cancer 4, 5, 34, 52, 53, 57, 61, 72, 73, 85, 110 breast 57, 72, 85, 110