New opportunities for improved personalized healthcare have emerged due to the recent advances in the development of modern methods which reinforce personalized early risk prediction, prevention and intervention. Using semantic techniques for data integration has become pivotal as it can deliver different ways to represent data, automating the process of data integration, and providing the ability to query semantically. In this paper, we propose a new semantic data model in which health information derived from Parkinson's, Multiple Sclerosis, and Stroke (PMSS) patients is systematically analyzed to generate and improve knowledge that will be transferred to patient care in order to design and develop innovative health risk prediction and intervention tools. Furthermore, this project focuses on providing new opportunities for improved personalized healthcare and prevention that have been created by new designs and developments of innovative health risk prediction and intervention tools. A core ontology is currently being designed within the ALAMEDA project to deal with the semantic interoperability across heterogeneous datasets along with a semantic framework to concrete the generated heterogeneous data through a shared ontology. The ontology model development and the requirement elicitation will be done based on the components' capabilities and use case requirements. The heterogeneous and dynamic data will be subjected to annotation through the development of semantic models for data sharing and usage apart from being interpretable.
Health-care ontologies are pivotal for knowledge representation and data integration as
the health data have become very complex and there is an intense need to link disorders
and applicable medication along with specific individual patient attributes so as to
extract meaningful results. The use of ontologies facilitates easier processing of large
datasets, while providing more effective solutions which support the way we manage
health and wellbeing and the indispensable integration of knowledge and data [
Semantic models are able to describe the health-care concepts and the relationships
between them by improving their effectiveness and efficiency. They have profited
healthcare communities with methods based on multiple ontologies, assuring the data
quality in a heterogeneous environment and organizing them so that it can be
interpreted by computers without human intervention. Using semantic techniques for data
integration has gained a lot of ground as they provide automated and multiple concepts
to represent and process the data, and allow for the semantically query [
In this paper, we present an Ontology based semantic model for health data interpretation, where the proposed model is able to harmonize information from multiple sources to provide context awareness. The use of the ontologies and the semantic web technologies will allow us to provide a conceptual model, supporting interoperability and flexibility. The ontologies will also address the semantic interoperability issues, management and integration of information models, as it will define the concepts and their relationships within the ALAMEDA’s domain. The project’s innovations will utilize new machine learning models, built upon lifestyle retrospective data as well as new streams of patient data that involve the monitoring of everyday activities, such as sleep behavior. The success of such applications will provide clinicians with the opportunity to modify interventions based on personalized data recordings. The main goals of this study are to implement the semantic models development for the annotation of the different modalities and to provide innovative solutions for context-aware data aggregation. For the purpose of this work, a health-care ontology is being designed to deal with health-care information and IoT devices and services.
The four steps based on Bravo, Reyes, Ortiz [
The development of semantic web technologies provides a number of ontology-based
approaches in different domains. Within the Semantic Web community, it is strongly
encouraged to reuse existing ontology models. The main domains that can be covered
are: Sensor Data, Context, Activity Recognition, Event and Healthcare ontologies.
Thus, we can build on existing resources and expand them, based on the specific needs
of the ALAMEDA project, where is necessary. There are several state-of-the-art
ontologies that can be utilized for modelling ALAMEDA’s domains. In particular, some of
the most commonly used ontologies, fused with healthcare interoperability standards,
are the Fast Healthcare Interoperability Resources-HL7 (FHIR-HL7) [
The Semantic Sensor Network ontology (SSN) [
Ontologies can be defined as an explicit specification of a conceptualization [
This study will provide the annotation layer to endorse ALAMEDA with situational awareness, extracting and harnessing deep intelligence through the aggregation of heterogeneous information and knowledge. It will also allow the comprehensive representation and the harmonization between the IoT devises, while it will offer innovative solutions development for context-aware data aggregation, enabling the semantic federation of diverse infrastructures and services for capturing information.
In this context, we are developing a health-care and patient-oriented ontology to personalize the medical and social knowledge available. The ontology will provide knowledge structures that are maintainable whilst they will be used to support clinicians in multiple tasks. The objective is to enable sensors and data coming from multiple distributed information sources to be semantically accessible and discoverable, fostering the development of data processing applications that effectively utilize and combine multiple data sources and devices to deliver innovative services. 4
In ALAMEDA project, we will design a consensual and conceptual model able: a) to represent information that is made available via the questionnaires and the monitoring modules, b) to ensure the semantic interoperability of the information exchanged between the individual ALAMEDA systems components, and c) to achieve the semantic annotation of the generated data and to further extend it with domain knowledge pertinent to the ALAMEDA use cases. It is crucial for the ALAMEDA model to represent information related to the domains of the project such as physiological and cognitive assessment, clinical data, reported difficulties etc. Within the framework of the ontology design, we reused the Even Model F ontology and the Event Ontology, so as to construct the ALAMEDA Event module. The Event Ontology models the events, the environment and the changes that may happen and answers a series of critical questions about the actions, the places, the time and the person of an event. During the Sensors Ontology construction, we reused the SSN Ontology, which as it has been already mentioned is a fundamental ontology in the sensor representation domain. 4.1
There are several methodologies for ontology engineering to formalise and design an
ontology. In this study, in order to develop the ALAMEDA’s Model, we used the NeOn
Methodology, which is based on a set of 9 scenarios and is well-documented and highly
adaptable. While building the ALAMEDA Ontology, different phases have been
determined. The first phase refers to the ontology requirements and the retrieval of the
ontology requirements specification document. The role of domain experts is very
important at this stage, since they define the use cases and propose optimal matching to
ontology requirements as the model is ongoing. The second phase is related to the
development of ontology at a primary level, where it will be defined which existing
ontologies will be used, along with the information input. The third phase contains the
implementation and enrichment of the ontology, using the OWL2 [
The identification of the purpose of ALAMEDA Ontology, the scenarios that must be defined, the intended uses and end-users are the main elements of the specification of requirements. As proposed in the NeOn Methodology, the methodological guidelines were created, based on the state-of-the-art ontology development techniques and represented in the Ontology Requirements Specification Document (ORSD). The requirements are defined with respect to the Competency Questions (CQs), which are groups of questions that play a significant role in the ontology implementation, as they specify what knowledge has to be entailed in. A critical attribute of the CQs is that they define the functional requirements of the ontology. The ORSD is the output of the ontology requirements specification and provides information regarding the a) purpose which is the main general goal and function that the ontology should fulfill, b) scope which refers to the coverage and the degree of details that the ontology should have, c) implementation language which is the formal language that will be used for the ontology design, d) intended end-users and uses and e) the non-functional and functional requirements. In ALAMEDA project, there are clinical aims of the use cases that are critical components for the ontology framework. Parkinson disease is a common neurodegenerative disorder and its meaningful worsening of global status or of individual motor is a specific use case with inclusion criteria such as advanced Parkinson, age and more. Relapse risk prediction in Multiple Sclerosis in young to middle age person is a second important use case. The Stroke use case refers to patients who suffered from Stroke the last month and is monitored for neuro-rehabilitation. The proposed Ontology should be constructed with respect to these use cases so as to provide a shared vocabulary for the communication and exchange of information among the different system components, to represent, store and retrieve patients’ profile data, sensors etc and to represent and query data made available by other analysis components of the ALAMEDA system. 4.3
The ORSD is a key factor for modelling the classes, properties and instances of the
ontology. Moreover, we used further input and feedback made available during the
stage of the ontology design by the clinical and technical partners involved in the
relevant project. Another element for the current version of the ontology was the standards
and best practices in the Semantic Web community which are available. The
implementation of the ontology was done in Protégé [
The ALAMEDA Ontology model consists of six modules and a main ontology, which
acts as the parent of all the hierarchical relations. The Model Ontology represents all
the modules attached to ALAMEDA, where Home, Person, Lab, Event, Time, Sensors
modules are some of the main classes. In Figure 1, a high-level figure of the model can
be seen.
Home Ontology formalizes information relevant to behavioural interpretation and
reported difficulties in the home environment. An example is the class
ReportedDifficulties which is used to describe information about the problems (e.g. difficulty in
exercising or bad mood etc) that a patient may face in the home environment. Lab Ontology
formalizes the types of information relevant to the tests, assessments, patient’s clinical
& experimental records in the lab environment. The class Domain is used to provide
information relevant to the specified domains, in order to describe the different types
of clinical tests and their results. The class MeasuredData indicates the data that is
essential to be shared during a task, while the class Task represents the possible types
of tasks involved in the ALAMEDA. Finally, the class ClinicalAssessment is utilized
so as to define the clinical characteristics that are collected during the clinical and
medical phases taking place in the lab environment. Person Ontology: refers to patient’s,
clinician’s and caregiver’s socialdemographic data. It consists of 5 classes that display
person, disease, gender, educational level and language. Event Ontology provides
information relevant to the entities and the activities that take place in order to fulfill
ALAMEDA purposes. The class Entity describes all the physical entities and consists
of 2 subclasses Person and Place. The class Activity represents any activity the patient
may be involved, while the class MeasurementPattern enacts the domain of the
measurements that take place. Event Model F played a crucial role in the development of
this module by using the participation pattern and making clear the roles and events in
our model, while the Event Ontology describes the time, the agents etc of the ontology.
Sensors Ontology describes information concerning the type and properties of the
sensors used which are divided into two major groups: a) FixedSensor and b)
WearableSensor. The class FixedSensor refers to sleep or location monitoring and the class
WearableSensor refers to sensors like smartwatches or belt sensors that will be used.
During the Sensors Ontology construction, we reused the SSN Ontology. The concepts
introduced by SSN are very important in healthcare sensing environment. Main
concepts that being reused are the class Procedure, which provides a way to specify
observation and has an input and an output. Those input and output information is
represented in the classes Input and Output. Other critical components of the SSN Ontology
that being reused are the Observation, the Platform. Time Ontology presents the time,
duration and information about the tasks of the ALAMEDA. It consists of classes like
DateTimeDescription, DayOfWeek etc that provide specific information about date,
time, day, duration and their values. Table 1 presents some of the most important
object/data properties of our model, their definition, the module/class they exist in and
one of their relations.
In Figure 2, an example of upper-level vocabulary for modelling a clinical test can be
observed [
One of the most challenging problems in healthcare systems is the interoperability between heterogeneous data, where a medium to share knowledge and exchange information both across people and services is essential. The Semantic Web services provide interoperability standards and vocabularies that can facilitate access to the necessary data in a secure and safe manner.
This paper describes a healthcare ontology-based model which interoperates between the systems and it is able to facilitate knowledge sharing in a complex environment. It is also able to manage and integrate patient specific data at home and lab environment with knowledge specific for this kind of patient. In addition, it presents the significance of this ontology, and provides users with the opportunity to gain insight and knowledge from their data and the criteria expected to be available through it.
Currently, the ontology development is ongoing, the requirement elicitation will be done based on components capabilities and use case requirements, while the heterogeneous and dynamic data will be subjected to annotation through the development of semantic models for data sharing and usage, besides being interpretable. An imminent challenge that needs to be addressed is to finalize the ORSD and the most appropriate semantic model for the acquired data that fits our case. In the future, we will include more restrictions and different types of properties on the interactions between intervention classes so as to implement a more efficient and accurate version of the ontology. Acknowledgements. This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No GA101017558.