=Paper=
{{Paper
|id=Vol-2656/paper4
|storemode=property
|title=Opportunities for Big Data Analytics in Healthcare Information Systems Development for Decision Support
|pdfUrl=https://ceur-ws.org/Vol-2656/paper4.pdf
|volume=Vol-2656
|authors=Blagoj Ristevski,Snezana Savoska,Natasha Blazheska-Tabakovska
}}
==Opportunities for Big Data Analytics in Healthcare Information Systems Development for Decision Support==
https://ceur-ws.org/Vol-2656/paper4.pdf
Opportunities for Big Data Analytics in Healthcare
Information Systems Development for Decision Support
Blagoj Ristevski, Snezana Savoska and Natasha Blazheska-Tabakovska
Faculty of Information and Communication Technologies, University St. Kliment Ohridski,
ul. Partizanska bb, Bitola, Republic of Macedonia
{blagoj.ristevski, snezana.savoska, natasa.tabakovska}@uklo.edu.mk
Abstract. Nowadays, an enormous volume of heterogeneous healthcare and medical
data are generated routinely. These heterogeneous data have to be integrated and
stored in a standard manner and format to perform appropriate big data analysis and
visualization and improve decision-making. These data are generated from different
sources such as mobile devices, sensors, national public health institutions, laboratory
tests, clinical notes, social media, and various omics data that can be structured,
semi-structured or unstructured. These data structure varieties necessitate these big
data to be stored not only in the relational databases but also in NoSQL databases.
To provide effective data analysis, besides the application of appropriate data
mining techniques, excellent design and implementation of healthcare information
systems are needed. These software solutions have to solve patient data security
and privacy issues by employing proper big data governance policies. The design
and implementation of healthcare information knowledge-based systems should
provide to the patients more well-organized and economical healthcare services,
and on the other hand, a boosted knowledge-based basis for decision-making to the
managers in healthcare institutions and insurance companies and benefits for the all
involved stakeholders. In this paper, we overview and suggest suitable development
framework that will cover patient-, clinical- and population oriented approaches
to decision-making and to reveal valuable knowledge and insights from these
healthcare and medical big data. Moreover, on specific occasions, this knowledge
should enable a rapid and reliable response to the healthcare hazards and help to
decision-makers worldwide as well on the national level.
Keywords: big data, healthcare information systems, knowledge-based systems,
decision-making framework, electronic health records.
1 Introduction
To obtain the optimal facilities and care for the patients, healthcare institutions
in many countries have suggested numerous models of healthcare information
systems. These models for personalized, predictive, preventive, patient-centric
Copyright © 2020 for this paper by
38 its authors. Use permitted under
Creative Commons License Attribution 4.0 International (CC BY 4.0).
and evidence-based medicine are based on using massive amounts of complex
biological, medical and healthcare data as well as electronic health records
(EHRs) [15]. As the final use, data are used for decision-making in healthcare and
medicine. Knowledge discovering from these medical and healthcare big data
allows identifying the best practices in the hospitals, discovering the association
rules and correlations in these data and unfolding the disease monitoring of
particular disease and patient- and population-centric health trends.
Nowadays, smartphones are excellent platforms to deliver personal messages
to patients to improve their welfare and health conditions and hence they are
crucial devices for the telemedicine, which is a very important branch of medicine
especially when special restrictive measures are established such as quarantines,
lockdowns and curfews.
To analyze and process health and medical big datasets that come from
different data sources, various techniques from many disciplines such as machine
learning, pattern recognition, expert systems, statistics, applied mathematics,
artificial intelligence are used. Many obstacles should be taken into account.
These big data are complex, often stored in distributed databases with medical
and healthcare records with a lack of integration capabilities and interoperability.
The development of new data mining techniques makes commonly used
machine learning algorithms easy to be adopted by bioinformaticians and to
become essential tools for the analysis of medical and healthcare big data. The
integration of these data provides a clearer picture of cell functions and alterations.
It will be more popular in the clinical health and disease examinations [16].
The new knowledge discovered by big data analytics techniques and
developed knowledge-based healthcare information systems should afford wide-
ranging and adequate advantages to the patients, clinicians, national public health
organizations and institutions, healthcare policymakers, as well as the World
Health Organization.
In this paper, we recommend directions for the development of knowledge-
based healthcare information systems considering several aspects and
functionalities.
The remainder of the paper is organized as follows. The concepts of big data
and big data analytics toward a framework for decision-making are described in the
second section. Section 3 describes the characteristics of healthcare information
systems. The next section explains the principles of knowledge-based healthcare
information systems in decision-making. The last section concludes this paper
with discussion and directions for further works.
39
2 Big data and big data analytics
Contemporarily high throughput bioinformatics technologies generate
large amounts of raw medical, biochemical and biomedical data, which are
heterogeneous like EHRs data, and stored in different data formats. Health and
medical big data refer to these numerous massive and complex data, which are
hard to analyze and manage with traditional software and hardware resources.
These big data can be categorized as structured, semi-structured or unstructured;
discrete or continuous data.
Big data analytics in healthcare and medicine covers merging of heterogeneous
data, control of data quality, analysis, modelling, visualization and validation.
Application of big data analytics provides thorough knowledge discovering from
the existing accessible large amounts of data [15]. Big data analytics in medicine
and healthcare has to enable analysis of large datasets from patients, when
public health of the whole population worldwide is important. Big data analytics
identifies patterns and clusters from available datasets, examine existing of a data
correlation, and develops predictive models using techniques from data science.
Another challenge when dealing with data mining techniques in big data
is the classification of an imbalanced dataset, which appears when real-world
applications produce classes with different distributions. One class is under-
presented with an insignificant number of instances, while the second one has
a plentiful number of instances. Identifying the minority classes is important in
various fields such as medical diagnosis, drug discovery or bioinformatics [22].
The rapid development of the emerging information and communication
technologies, experimental technologies and methods, cloud computing, the
Internet of Things (IoT) and social networks provides the amounts of generated
data that grow massively in medicine and healthcare as well as in other domains.
These high throughput data, that are so-called omics data, provide widespread
insights towards different types of profiles, changes and interactions on a molecular
and cellular level as well as knowledge associated to the genome, epigenome,
transcriptome, proteome, metabolome, interactome, exposome, diseasome, etc.
Besides these omics data, the EHRs data contain personal patients’ data, clinical
notes, diagnoses, administrative data, prescriptions [18], as well as charts, reports,
laboratory tests, medical images, magnetic resonance imaging, ultrasound,
tomography, X-ray data. Some of these data are acquired from wearable sensors
or capture from medical monitoring devices, with different collection frequency
that makes these data to have complex features and high dimensionality [15].
These growing amounts of various omics and EHRs data need to be collected,
cleaned, stored, transformed, transferred and visualized in an appropriate fashion
to be represented to the clinicians and healthcare organizations and institutions.
40
Thus, the development of appropriate healthcare information systems that are
based on knowledge and big data is essential.
The term big data is usually described by the following features: volume,
value, velocity, variety, veracity and variability, which are denoted as 6 “V’s”
big data characteristics [15]. The volume of healthcare and medical data,
usually measured in terabytes, petabytes and yottabytes, refers to the quantity
of data, while value ascribes to the comprehensible and valuable data analysis.
Velocity is associated with the shifting data in motion as well as and to the
speed and frequency of their creation, processing and analysis. Complexity and
heterogeneity of multiple datasets refer to the data variety. Veracity referrers to
the quality, relevance, uncertainty, reliability and predictive value of the data,
whereas variability is linked to the data consistency for a while.
Applications of big data analytics can improve the patient-based services, to
detect spreading diseases earlier, generate new insights into disease mechanisms,
monitor the quality of the medical and healthcare institutions as well as provide
better treatment methods, especially for novel infectious diseases, when treatment
techniques update often. If a disease occurs and cures in any part of the world,
then prediction and modelling for that disease can be done competently when
using big data analytics.
Big data analytics has the potential to transform the manner healthcare
providers use advanced technologies to obtain knowledge from clinical and other
data repositories and hence to make better decisions [12].
3 Healthcare information system
The biggest challenge today related to the analysis of big data in medicine and
healthcare is the big data integration from many data sources that generate large
amounts of data. Unfortunately, there is no pre-defined strategy to the healthcare
data integration. Although there are many ontologies connected with healthcare
and medicine as well as the factors that affect human health that attempt to
integrate data on the principle of data “born interoperable”, as shown in Fig. 1.
Some health data are well structured according to well-known medical and health
coding systems but many of them do not have a predicted structure that could
allow the ontological big data interoperability. Many commonly used ontologies,
such as Gene Expression Ontology, Gene Ontology, microRNA Ontology, Protein
Ontology, MONDO disease Ontology, Disease Ontology etc., as shown in Fig. 1,
consider the standards established in healthcare and medicine and introduce well-
chosen indicators that should help to reveal the hidden links between the data.
Most medical and health data use already known coding systems in medicine,
which is one-step towards improving the big data analysis. A well-known and
good instance of this is the ontologically enabled big data integration into
41
toxicology [4] that uses existing efforts and ontologies to link data using the
knowledge graph to gain new knowledge. A knowledge graph is a tool used also
in many domains. As an instance, Google’s search algorithms since 2010 and the
improved version of integration with the using of ontology since 2018 enable
much more pervasive searching algorithms that allow finding hidden connections
between data [5].
Another good example of data integration is the connections of biomedical
data throughout most known biomedical ontologies that integrate data obtained
experimentally in wet labs and reuse them via established ontological systems
with predefined metadata that are more widely available. This integration of
healthcare, omics, exposure and medical data is only part of the mosaic of total
data sets used in medicine and healthcare, shown in Fig. 1.
The data, collected in healthcare institutions, laboratories, national and other
health-related systems, should be added. Considering a huge amount of data
collected from many sensors, widely used by many patients, such as Holter and
other measuring sensors that show vital life signs, according to the principle of
smart living with IoT, the data science techniques need to be employed to deal with
these big data. In addition, the new trend of using environmental data that affect
human health provides a wide range of applications in healthcare and medicine.
These data are related to systems from various measuring stations for scientific
research in various domains, such as measuring the number of nanoparticles,
electromagnetic radiation, the concentration of pollutants in the stratosphere,
chemicals that affect health and many others sensors that measure environmental
pollution, radiation, various types of soil, water but also human nutrition data.
42
Fig. 1. A part of the mosaic of ontologies of whole data sets used in ecosystems in medicine and
healthcare.
Today, the largest aggregators of healthcare data are smartphones with
many applications used by users. They also do not have pre-defined structures
and measurement methods but follow well-known and accepted healthcare
standards. Linking of all these data with the need to risk assessment connected
with genotype, phenotype and disease is another branch that has been intensively
exploited since 2006 as exposome data. They are related to the impact of external
factors on the genotype and phenotype and intend to provide the risk calculation
43
for each individual based on data from his/her EHR/PHR and outer influences
of the existing big data repositories for various factors. Also using artificial
intelligence methods to quantify location-based risk and finding an influence of
all previously mentioned data is required [10] [19].
Another important challenge includes links between the patient-based
database (EHR), e-health data and all data that allow evidence-based medicine,
pervasive healthcare and telemedicine [7]. The continuous migration and
movement of people require from patient to bring their healthcare data with
themselves when travel. Unfortunately, these data are not usually accessible for
patients because data are owned by institutions that provide healthcare. So, the
patients when are abroad and physicians who have to make decisions do not have
enough data to practice evidence-based medicine. It suggests that medical facts
are needed at every level of decision-making, especially for the treatment and
care of patients for evidence-based medicine [1]. Although these data are stored
in EHR, they are not available outside the institutions and the country where
the patient is a citizen. Therefore, many efforts are aimed toward supporting the
creation of PHR for patients and allow the patients to have access to all data
related to their health statuses such as EHR from medical practitioners, institutions
visits, medical research [10], laboratory and biometric results, prescriptions and
referrals and other data related to the screening of patient’s health. Availability
and accessibility of such data are activities towards improving evidence-based
medicine and making good decisions by physicians to whom the patient addresses.
If the patients have their health data, the concept is PHR allowing evidence-based
decision-making and medical treatment when the patients are abroad.
This concept requires using web-enabled technology with e-health and
evidence-based PHR, which has many advantages. For physicians, adopting
PHR is important because it allows improving the decision-making process from
everywhere, based on evidence and for the patient to receive better evidence-based
diagnosis and treatment from medical practitioners. IoT based health-monitoring
sensors included as wearable measurement sensors also improve evidence-based
decision-making. These devices connected with Bluetooth can capture and store
health-related data. PHR can also contain data obtained from the equipment such
as accelerometers, gyroscopes, wristband and smartwatches usually connected to
smartphones and stored in PHR via Bluetooth connection [14].
Nowadays, when we are facing epidemics of various diseases associated with
many autoimmune diseases that require genetic analysis but also pandemics that
requires analyzing of genetic data of many microorganisms, bacteria and viruses,
we are aware that data integration is an extremely important task. Patient-centric
decision-making should give way to different analysis of patient groups and high
dimensional data related to diseases, behaviors, treatments and many omics-data.
This requires the application of global data interoperability standards such as
44
those for PHR, proposed by ISO in 2012, introduced as EHR-ISO/TR 14292 and
applied as HL7 standard [13]. Because these data have very variable formats and
often they protected by private data protection standards and laws, they cannot
always be used for analysis. Data decoding and aggregation are sometimes
required to enable data analysis such as population-centric, epidemic-centric,
clinic-disease-centric, hospital-centric, region-centric [25], country-centric, to
support decision-making by healthcare stakeholders at different layers, as shown
in Fig. 2.
Fig. 2. Benefits from big data analysis for decision-making in three layers.
When benefits from big data analysis in healthcare and medicine are taken
into account, a couple of aspects have to be considered. The first aspect, clinical
conclusions as historical reports, statistical analyses as well as time-series analyses
and comprehensive reports have to provide evidence-based decision making in
medicine and healthcare, especially for diagnostics and healthcare treatments
[24]. The second aspect is the information visualization that enables interpolation
of critical big data for analysis using interactive dashboards or charts that support
daily operations of physicians and nurses and helps them to make more efficient
and faster decisions [17]. The third aspect is real-time reporting as warnings,
alerts and proactive notifications, real-time navigation and the application of
operational key performance indicators that are usually placed on dashboards in
real-time.
According to the framework suggested by Shang et al. [21], there are
45
five benefit dimensions, which include IT infrastructure benefits, operational
benefits, organizational benefits, managerial benefits, and strategic benefits.
This framework is suitable for a more general system model for categorizing
the benefits of big data analytics. They consider the content analysis that takes
place in a three-phase process: preparation, organizing and reporting. It provides
a better comprehension of big data analytics capabilities and healthcare benefits.
They have identified 5 big data opportunities: analytics capabilities for big data
analytics, insights capability, predictive analytics capability, interoperability
capability and traceability.
4 Using knowledge-based healthcare information systems in
decision-making
The success of an organization depends on the knowledge quality. To support
knowledge management in healthcare organizations, health information systems
must provide information and guidance to the medical personnel and patients
and help them in decision-making. Health information systems (IS) are complex.
They cover a wide and diverse range of applications, including hospital IS,
nursing IS, laboratory IS, radiological IS, pharmaceutical IS, EHR systems,
patient monitoring systems, clinical decision support systems, medical education
system etc.
Additionally, healthcare decision-making has to be a knowledge-driven
process, so knowledge management tools in the healthcare sector are very
important. Providing the right knowledge at the right time at the point of decision-
making by implementing knowledge management in healthcare is crucial.
The typical architecture of a knowledge-based healthcare information system
includes a knowledge base and an inference engine. Health information system
backed by a rich and effective knowledge base, which contains a collection of
information of the field of medical diagnosis, ensures efficiency in identification,
analysis, and selection of optimal action for the patient care. Inter-organizational
knowledge sharing is one of the fundamental steps in knowledge management
processes and can serve as a strategic system for knowledge-intensive sectors
such as healthcare [20]. The inference engine deduces insights from the
information stored in the knowledge base. This improves access to the patient
data and facilitates the decision-making process in the shortest possible time.
The healthcare person through the interface interacts with the system during the
decision-making process.
The knowledge-based healthcare information systems use appropriate
tools for knowledge management and user-friendly interactions because they
can significantly improve the quality and safety of care provided for patients
at the hospital and home surroundings [20]. They assist in the data collection,
46
analysis, management and sharing of knowledge between business processes for
healthcare. The vital aspects of knowledge-based healthcare information systems
are utilization, transfer and translation of knowledge. Knowledge utilization is the
process of converting knowledge, such as evidence-based guidelines to practices.
Knowledge translation moves scientific knowledge from basic discovery to
testing for technical efficiency and then to acceptability for adoption in practices.
The third aspect, knowledge transfer, is the diffusion of knowledge that is directed
and managed by using various strategies [20].
The healthcare providers play a significant role in patient care such as
enhancing the quality of care, ensuring individual based on the most up-to-date
evidence, ensuring physicians to maximize the likelihood of positive outcomes
as well as minimizing the existing gap between research and practice [2]. The
Boateng [2] and Lapaige [8] defined four steps in decision-making and actions
taken in the healthcare industry: 1) formulation of a clear clinical question related
to the patient problem; 2) a search in literature for relevant clinical practices; 3)
the evaluation of the available evidence for its usability and 4) implementation of
the evidence in clinical practices.
There are many currently existing knowledge management tools to implement
in healthcare organizations. These tools are interactive, most initiated and allow
effective communication between healthcare professionals, managing their
knowledge, generating discussion about new concepts or ideas, finding answers
to particular problems.
Other opportunities for healthcare professionals include clinical decision
support systems, EHR system, the community of practices and advanced care
management. To obtain a full-fledged implementation of knowledge in healthcare,
all stakeholders such as policymakers, researchers, health professionals and
healthcare providers need to come together and play their part to seize the
opportunities and improve healthcare quality [20].
Furthermore, public health practitioners need to be fully informed when
deciding on the design of an information system, integration and using of data [3].
Accordingly, the knowledge-based information system is expected to play more
roles in healthcare in the future. The design of knowledge-based information
system in healthcare organizations are gaining more attention. The most important
and challenging task in designing a knowledge-based information system is to
organize and maintain patient information repositories securely, accurately and
in a speedy manner [18].
As knowledge-based information systems’ applications increase further in
the healthcare sector, some privacy and security issues will arise, as well as issues
related to the data transparency, drug prescription ad supply chain errors, integrity,
accessibility, resource and patient management and knowledge interpretation [9].
Over time, all of the health information will be available electronically, to the
47
patient, to the doctor and other healthcare providers. Because many organizations
and people may have access to health information, there will be concerns about
the privacy and security of health information [11].
Because big data flood in medicine, healthcare, and the nature of medical and
healthcare data, new database management systems such as Cassandra, MongoDB,
MarkLogic and Apache HBase and NoSQL database systems should be employed
for the development of knowledge-based healthcare information systems [23].
The framework of developed IS should enable data analysis on a different tier,
such as patient-centric, population-centric, epidemic-centric, clinical-centric,
country-centric, to support decision making by the heads of the healthcare and
medical institutions and organizations. Furthermore, data privacy and security
issues of the patient sensitive data have to be solved in every proposed software
solution by using appropriate anonymization and cryptographic protocols.
5 Conclusion and further work
The healthcare industry generates many Exabytes healthcare data, manly in the
form of EHR. However, most of the achievable values of data usage do not obtain
full potential. This process is still in its infancy because predictive modelling and
simulation techniques for analyzing healthcare data as a whole have not yet been
adequately developed [23]. Big data analytics in medicine and healthcare offers
very promising possibilities for the development of decision support systems
and knowledge-based healthcare information systems that will integrate explore
and analyze large amounts of data. These integrated information systems should
make a symbiosis among patient-, clinical- and population-centric decision-
making systems [6]. As further work, the big data characteristics provide a very
appropriate foundation to use promising software platforms for the development
of applications that can deal with healthcare and medical big data. Moreover,
the development of healthcare information systems has to consider solving the
security and privacy issues of all involved parties, especially patients’ sensitive
data, and these software solutions have to enable patient-centric, population-
centric, epidemic-centric, clinical-centric, country-centric data analysis to make
decision making more effective.
48
References
Andreu-Perez J, Poon CC, Merrifield RD, Wong ST, Yang GZ. Big data for health. IEEE jour-
nal of biomedical and health informatics. 2015 Jul 10;19(4):1193-208.
Boateng W. Knowledge management in evidence-based medical practice: does the patient mat-
ter?. Electronic Journal of Knowledge Management. 2010 Nov 1;8(3).
Biancone P, Secinaro S, Brescia V. A review of big data quality and an assessment method
and features of data quality for public health information systems, International Journal of
Management Sciences and Business Research, Jan-2018 ISSN (2226-8235) Vol-7,
Issue 1
Boyles RR, Thessen AE, Waldrop A, Haendel MA. Ontology-based data integration for ad-
vancing toxicological knowledge. Current Opinion in Toxicology. 2019 Aug 1;16:67-74.
Chah N. OK Google, What Is Your Ontology? Or: Exploring Freebase Classification to Under-
stand Google’s Knowledge Graph. arXiv preprint arXiv:1805.03885. 2018 May 10.
Eapen BR, Archer N, Sartipi K. Public Health Information Systems: From Data to Knowledge,
2020.
Househ and all, Big Data, Big Challenges: A Healthcare Perspective Background, Issues, Solu-
tions and Research Directions, Fernando Martin-Sanchez, Big Data Challenges from an Inte-
grative Exposome/Expotype Perspective, Springer 2019, ISSN 2195-271X, ISSN 2195-2728.
Lapaige V. Evidence-based decision-making within the context of globalization: A “Why–
What–How” for leaders and managers of health care organizations. Risk management and
healthcare policy. 2009;2:35.
Litchfield AT, Khan A. A Review of Issues in Healthcare Information Management Systems and
Blockchain Solutions. CONF-IRM. 2019.
López-Campos G, Merolli M, Martín-Sánchez F. Biomedical Informatics and the Digital Com-
ponent of the Exposome. InMedInfo 2017 (pp. 496-500).
Mahmood N, Burney A, Abbas Z, Rizwan K. Data and knowledge management in design-
ing healthcare information systems. International Journal of Computer Applications. 2012 Jan
1;50(2):34-9.
Milenkovic MJ, Vukmirovic A, Milenkovic D. Big data analytics in the health sector: challeng-
es and potentials. Management: Journal of Sustainable Business and Management Solutions in
Emerging Economies. 2019 Mar 19;24(1):23-33.
Morgenthaler J. Moving Toward an Open Standard, Universal Health Record. Smart-publica-
tions. 2007.
Rappaport SM, Smith MT. Environment and disease risks. science. 2010 Oct 22;330(6003):460-
1.
Ristevski B, Chen M. Big data analytics in medicine and healthcare. Journal of integrative
bioinformatics. 2018 May 10;15(3).
Ristevski B, Savoska S, Mitrevski P. Complex Network Analysis and Big Omics Data, ICT In-
novations 2019, RN of Macedonia, 2019.
Roski J, Bo-Linn GW, Andrews TA. Creating value in health care through big data: opportuni-
ties and policy implications. Health affairs. 2014 Jul 1;33(7):1115-22.
Sarkar BK. Big data for secure healthcare system: a conceptual design. Complex & Intelligent
Systems. 2017 Jun 1;3(2):133-51.
Savoska S, Ristevski B, Blazheska-Tabakovska N, Jolevski I. Towards Integration Exposome
Data and Personal Health Records in the Age of IoT, ICT Innovations 2019, RN of Macedonia,
2019.
49
Shahmoradi L, Safadari R, Jimma W. Knowledge management implementation and the tools
utilized in healthcare for evidence-based decision making: a systematic review. Ethiopian jour-
nal of health sciences. 2017;27(5):541-58.
Shang S, Seddon PB. Assessing and managing the benefits of enterprise systems: the business
manager’s perspective. Information systems journal. 2002 Oct;12(4):271-99.
Oussous A, Benjelloun FZ, Lahcen AA, Belfkih S. Big Data technologies: A survey. Journal of
King Saud University-Computer and Information Sciences. 2018 Oct 1;30(4):431-48.
Wang Y, Kung L, Byrd TA. Big data analytics: Understanding its capabilities and potential
benefits for healthcare organizations. Technological Forecasting and Social Change. 2018 Jan
1;126:3-13.
Wang Y, Hajli N. Exploring the path to big data analytics success in healthcare. Journal of Busi-
ness Research. 2017 Jan 1;70:287-99.
https://ec.europa.eu/health/indicators_data/data_en
50