Health care is one of the popular segments wherein a huge amount of data is being generated every second and in a heterogeneous format. It is only the use of semantic technologies that can act as a guard for data integration and enriching it by adding proper semantics. The Semantic Web gives us a sensible and versatile answer to perceive proficiency across the congregation of healthcare data. Eventually the ultimate goal of improving the health care practices depends on how the healthcare data is linked together. The major challenges to chase the goal are not only limited to enable the integration of heterogeneous data sources but also requires the development of certain tools for efficient search, ontology management, and data analytics. The major objective of this study is to understand the technologies behind the Semantic Web and to unshield the mysteries of application of Semantic Web in the healthcare segment. The data for such an extensive survey was gathered from repositories of Pub Med, Scopus, Web of Science, and Google Scholar along with few medical science libraries.
ACI’21: Workshop on Advances in Computational Intelligence at ISIC 2021, February 25-27, 2021, Delhi, India EMAIL: pallavihptu@gmail.com (P.Nagpal); deepika.chaudhary@chitkara.edu.in (D.Chaudhary.); jaiteg.singh@chitkara.edu.in (J.Singh).
2020 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR Workshop Proceedings (CEUR-WS.org)
The better definition of Semantic Web (SW) always begins with the word Semantic. In simple words semantics signifies meaning. Meaning allows effective use of data. Meaning is mostly missing in many of the information sources and is the job of the user or programming instruction to deliver the same. For example, if we simply look at the <H1> tag it is used to highlight major headings but as technical people, it is understood that the text which is surrounded by <H1> tag is of more importance to the user when compared to other. Basic semantics are attached for the search engines with <META> tags; however, they are just solitary keywords and lacks when linked. Semantics accord keywords with effective meaning through relationships. The Semantic Web in simple terms is a web of data paraphrased and linked together in ways to build context that follow a grammar & language construct. Although these semantics can also be added with a help of programming language in that case also formal standards are missing and also these semantics cannot be shared, aggregated, and validated. One can also attach these semantics through programming languages but in that case, too, there are no formal standards to be followed, and sharing, aggregation, and validation among statements becomes complex if not impossible. The Semantic Web all together supports the evolutionary nature of WWW. The Semantic Web architecture is based on the Layered Approach. These layers have some strong key dependencies between them. The Semantic Web layer cake presented in Figure 1 highlights these dependencies.
Identifier (URI) – This layer uniquely identifies the available resources on the Internet. This layer ensures the uniqueness of Objects.
Language (XML) – A language that lets it underlying user to create web pages with a defined vocabulary. This layer also supports sharing documents across the web.
Description Framework (RDF) - RDF acts like an Entity-Relationship(ER) model which is drafted for writing a set of statements to describe objects (resources) available on the web. RDF Schema on the other hand is used to write ontologies to represent relationships between web objects [26]. For integrating richness to the interoperability of data among various domains Ontology Web Language (OWL) was developed.
layer is all about the representation of logics to deduce one document from another and there validations. The trust layer ensures the faith through the use of digital signatures.
To conclude we can state that the Semantic Web can be better understood as an extension of the WWW, it extends itself through the usable, systematized semantics that is based on research in knowledge/facts representation and the logic to approach the goal of ubiquitous information sharing. When compared to WWW it's the Semantic Web which primarily consists of statements for application consumption. These statements are not just plain statements for human interpretations but also include logic and act as meaningful links which can directly be interpreted by machines.
The outcome of this paper is to provide insights on the recent advancements done in this field. The more specific objectives can be summarized as follows:
a) To create a theoretical foundation in the area of data standards and interoperability, ontologies and its visualization, semantic data repositories and Semantic Web user Interfaces.
b) To reveal the current state of art of application of Semantic Web in Health Care Domain.
c) To un-shield the successful use cases of new approaches, techniques and Semantic Web applications in the field of health care data management.
The next section presents the present day challenges which require attention of the researcher community.
The use of information systems and technologically advanced medical devices in various health care hospitals is producing and will continue to produce vast amount of data. This huge collection needs to be explored and transformed in such a way that it can be converted into valuable information which can thus improve the healthcare processes. Which is not easy and can be daunting because of various challenges? The following challenges needs to be addressed:
a) Interoperability of health and medical data.
b) Personalization of Ontologies and its visualization.
c) Explosion of health data and Semantic Data Repositories
d) Development of new user friendly interfaces
Semantic Web is attaining approval in addressing these challenges and therefore World Wide Web consortium (WWWC) established the Health Care and Life Science Interest Group in the year 2015 (HCLSIG,2015) with the objective to defend and support the use of Semantic Web technologies in health care segment. The rest part of the paper is organized as follows section 3 provides a background and short outline of prevailing literature for the mentioned challenges, section 4 highlights the new techniques and approaches relevant to health care segment, section 5 concludes the study.
This section presents a short outline of a few of the prevailing literature towards Semantic Web technologies in the health care segment. In this phase, a detailed study of literature has been carried on to improve the quality and safety of web applications in the health care domain.
The automation of various health care services; usage of various medical information systems and other technological instruments have put together and still is generating volumes of medical data and that too in heterogeneous form. The huge volume of data has to be processed in a meaningful form to produce some valuable knowledge. This knowledge can lead to sound health care practices and thus can be of utmost use for humans[26]. When the data is coming from multiple sources it's the interoperability standards that act as a bridge to integrate and exchange the data across systems and services. To achieve this requires Schema matching techniques that can transform the data from human-understandable to machine-ready format. This technique when implemented can produce good healthcare services thereby reducing the cost by eliminating duplicate operations. The authors in their work have highlighted the meaning of interoperability for Semantic Web applications which can be defined as the vocabulary, organization, and structure of data required to integrate the data from multiple institutions [28]. Figure 2 gives a glimpse of various standards dealing with interoperability and the changes concerning time. The interoperability can be categorized into two parts a) functional and b) semantic where functional deals with the common procedures and semantic deals with the framing of a common language which machine can understand in the end to end communication. To make applications fully interoperable requires standards. As per the literature mentioned the standards can be divided into a) Vocabulary or ontology Standards b) Data interchange and Integration standards c) Health record maintenance standards. These standards are established through four methods a) Adhoc b) Defacto b) Governmentmandated) Consensus.
Below we present the literature survey done in this direction:
The authors worked on schema matching
techniques and presented a method for
automating the process for matching schema at
the field level; they achieved 71.8% when
mapping the four staged process which
includes string matching and substring
matching [25]. The authors in their paper have
discussed the standards required in Clinical
data management [15]. Clinical data
management deals with signs, operations,
medicines, and lab values for a particular
patient. The data here can only be
interoperable when in a structured form. This
transformation is a challenge and requires a lot
of research. Further data analysis also requires
the data to be converted in a structured form.
The authors have given the FAIR guiding
principle as one of the possible solutions to
this problem. The acronym of FAIR is
"Findability", "Accessibility",
"Interoperability", and "Reusability". The
authors in the study have mentioned Fast
Healthcare Interoperability Resources (FHIR)
as a newbie in the area of biomedical
informatics and healthcare [26]. These
guidelines when followed can provide a
technological edge over health level seven
(HL7). It was also discussed in the Yosemitea
manifesto (https://www.dataversity.net/
semantic-interoperability-future-healthcaredata/) that for data exchange the Resource
Description Framework (RDF) to be used.
Existing data standards to be mapped with
RDF's, Government agencies should mandate
RDF as a Universal Healthcare exchange
language. These standards can be adopted in
the area of legacy systems, modern medical
entities, and healthcare information systems.
This standard can also leverage RESTFUL
web services. The given section categories the
literature review underwent by various
researchers.
1. Private and public sector organization to
align their systems with Fast Healthcare
Interoperability Resources is the upcoming
standard for data exchange.
(https://www.dataversity.net/
semanticinteroperability-future-healthcare-data/ ).s
[14]
2. Development of various standards for
providing data interoperability; XML,
GPS, Web Services, and Security, TCP/IP,
802.11, GPS [
Ontologies play a very important role in Semantic Web applications. They act as a common vocabulary for a specific domain consisting of a terminological collection of terms together with the rules to combine these terms and form relations. They also act as a basis for interoperability.
Here, we first survey the work done in this direction. The authors in this paper did an extensive survey on the current state of ontology development in the area of health care [20]. They have studied various papers in the time frame of 2009 and 2018. They classified the work done by various researchers and proved that at present also there is a dire need for the development of new ontologies for delivering effective healthcare services. The authors in this paper have highlighted the role of ontology in decision making and how ontology that defines concepts such as disease, location, and environment and the interrelationships can influence the process of decision making in the public sector domain [26]. They also focused on the various mapping methods to work on interoperability issues. Further, they have developed a centralized knowledge base for healthcare systems specifically for the Tamil Nadu Region of India. The author in this paper has integrated the data which is coming from different sources and thus created an ontology [19]. They have followed four-way steps which include identification of data sources, the formalization of concepts, performed audit, and thus formulated ontology. In the table given below we summarize the work done in the direction of creating new ontologies into past, Work in progress, and Future Scope. The authors stated that today also lot of data which is published by big organization like WHO is published in proprietary format and not in accordance with Semantic Web standards and therefore it is still very difficult to integrate and further process those chunks [28]. Although a lot of research work (WHOGO, 2015), (PubMed, 2015), (NIH, 2015) has been carried in this direction still there is a long way ahead. The table below summarizes the literature. 1. Development of a Centralized knowledge base. Personalization of ontologies for describing the class hierarchies among chronically ill patients to form a decision support system for chronically ill patients (Riaño et al., 2012) 2. Introduction of Electronic Health care system to improve the various parameters of health care services [19]. 3. Improvement was done in various organizations dealing in the service delivery segment, availability of reliable health data for healthcare providers, and improves upon in the public health system. A lot of new ontologies were also developed. The process of ontology validation was also improved [27]. 4. Development of comprehensive monitoring frameworks in the field of maternal health informatics that would be created with the consensus of people practicing it and also on the ontologybased data integration approach. These frameworks will facilitate the research and evidence-based decision support systems [11]. 5. To work upon improving the quality of health services using semantic web technologies. Majorly the privacy and security, trust, risks, and social implications and the quality of information are important and play an important role in the semantic web areas [15]. 6. HCLS Knowledge Base is designed as a knowledge base where data from multiple sources (PubMed, Clinical Trials) have been stored. WHOsGHO – From WHO statistics about 3 million of data has been converted to RDF [23]. Other projects similar to the one mentioned above like DailyMed (2015) would continue to generate data from the health domain based on the principles of linked data [23] 7. The data hub is currently indexing hundred of datasets which are tagged under the healthcare category. The requirement of a lot of new projects in this direction which can convert the data stored in a proprietary format to linked data [23]
3.3.1 Theoretical Foundation There are multiple projects such as CardioShare and Bio2RDF which have certain capabilities for navigating and querying the underlying Semantic Web data. However they lag intuition and can be more improvised.
The projects which are designed in this
direction should have the capabilities for
searching and navigating through the Semantic
Web data [28]. There are certain projects
which have these capabilities like Bio2RDF
and CardioShare, but these projects are limited
and are not that intuitive which means a novice
user will find it very difficult to explore and
visualize the RDF triples. Therefore, there is a
strong need to develop a good interface while
developing the Semantic Web Applications.
The architecture is so simple that even a
beginner can explore it without much
difficulty. The authors in this paper have
designed and Semantic Web Portal (SWP)
which is a light weight portal to browse and
visualize the data generated and that too in
meaningful and friendly way [
Although the technologies in the underlying domain have contributed a lot in health care segment still this study reveals that there still certain challenges that needs to be resolved. This survey reveals the present day challenges which require an ultimate attention of the researcher .The end users can be benefitted in case we get an optimal solution to the challenges mentioned herein.
Healthcare segment is generating a lot of data continuously. Semantic interoperability of data can help the human community a lot. The patients, the doctors, various organizations can get the maximum benefits from this. The figure below represents the future issues that need to be addressed to make this technology a huge success. a) Semantic Data Integration for IoT Sensor Data
In present scenario, about 35 billion of IoT
devices are connected and it is predicted that
this number would grow around 120 billion in
2025, which would be generating around 180
trillion gigabytes of data. This data comes
from various heterogeneous devices thus
making the formats incompatible to integrate.
Which creates a significant problem for IoT
application developers? Semantic Annotations
and Clustering can be used as a method to
integrate this data which is a challenge and can
be considered as a future scope [
Irrespective of the complexity and size of
data interfaces act as central linkage between
human computer interactions. This becomes
more complex when data is coming from
various heterogeneous systems. Therefore,
there is a need of designing new faceted and
interactive interfaces [24]
c) Integration of Machine learning
algorithms into Semantic web reasoners
Certain new reasoners are required in the area
of the health care domain to generate
inferences that can learn from themselves and
are based on ANN or deep learning techniques
[