CC BY
25
w
AN ONTOLOGICAL-BASED MONITORING SYSTEM FOR PATIENTS WITH BIPOLAR I DISORDER
Chryssa H. THERMOLIA1, Ekaterini S. BEI1, Euripides G. M. PETRAKIS1, Vangelis KRITSOTAKIS2, Vangelis SAKKALIS2
1 School of Electronic and Computer Engineering, Technical University of Crete 73100, Greece, Chania, Akrotiri Campus
2 Institute of Computer Science, Foundation for Research and Technology 71110, Greece, Heraklion, N. Plastira, 100
Our aim is to provide a patient monitoring system that integrates a Clinical Decision Support System (CDSS) and an Electronic Health Record (EHR) that assist psychiatrists and primary care physicians to tackle existent health needs of mental illness related to the treatment and management of bipolar I disorder (BDI). Our monitoring system consists of an EHR system based on the Health Level Seven Reference Information Model (HL7-RIM) and an ontological-based CDSS leveraging the Semantic Web capabilities. Based on the evidence-based clinical guidelines and patients' health records, the monitoring system is developed to encode and process this information and subsequently to assign recommendations of choices and alerts to clinicians for improved mental health care. Considering the clinical guidelines germane knowledge, as well as issues of patient's health record, the monitoring system can support a personalized decision-making for bipolar I disorder longitudinal course. We propose AI-CARE as an online monitoring tool that may offer useful guidance in clinical practice.
Keywords: clinical decision support system, electronic health record, semantic web, ontology, bipolar disorder.
Optimal health care is a core challenge of several emerging technologies in order to promote the best health care conditions and improved health outcomes for patients living with chronic diseases, such as bipolar disorder. Toward this challenge, the advent of new scientific discoveries in medicine (genetics, epigenetics, pharmaceuticals) along with the technological explosion (medical devices, internet) enable the development of computer-based monitoring systems to aid clinicians in promoting high-quality health care and meeting the goals of «personalized medicine» [10, 12]. Personalized medicine aims to the effective adaptation of biomedical and technological knowledge to the individual features (genetic, anatomical, and physiological characteristics), needs and preferences of each patient mainly considering the difference in patient's susceptibility to a specific disease or patient's response to a particular medical therapy; which does not literally translate in the production of unique drugs or
intelligent devices for a specific patient but rather a more targeted therapeutic intervention [12].
In the new era of personalized medicine is highlighted the need to combine «evidence-based medicine» with case based reasoning in order to enhance the health care process [4]. Evidence-based medicine refers to «the conscientious, explicit, and judicious use of current best evidence in making decisions about the care of individual patients», while its practice implies the «integrating individual clinical expertise with the best available external clinical evidence from systematic research» [14]. In this context, CDSS systems are favorable tools to promote the practice of evidence-based medicine and clinical guidelines, in turn is a common method for CDSS [4]. These types of CDS systems, which include documented clinical knowledge, are called knowledge-based systems and provide guidance to clinical decision making [2].
Thermolia Ch.H. - student, the Intelligent Systems Laboratory, e-mail: cthermolia@ intelligence.tuc.gr Bei E.S. - Ph.D., postdoctoral researcher, the Intelligent Systems Laboratory, e-mail: abei@isc.tuc.gr Petrakis E.G.M. - Ph.D., Professor, laboratory director, the Intelligent Systems Laboratory, e-mail: petrakis@intelligence.tuc.gr
Kritsotakis V. -M.Sc., technical staff, Computational BioMedicine Laboratory, e-mail: vkrits@ics.forth.gr Sakkalis V. - Ph.D., principal researcher, Computational BioMedicine Laboratory, e-mail: sakkalis@ics.forth.gr
In order to interpret information and derive knowledge we use Semantic Web Technologies. The concept of ontology is basic element of Semantic Web, defining a set of primitives containing [11]: concepts, relationships between concepts, described by domain and range restrictions, the taxonomy of concepts with multiple inheritance, axioms describing additional constraints on the ontology that allow to infer new facts from explicitly stated one. Semantic Web Technologies also includes the application of sets of rules that enable us to model knowledge, by inferring new implicit axioms, based on explicitly specified ones, checking for satisfiability of classes, computing hierarchies of classes and properties and checking consistency of the entire ontology.
In our implementation the existence of a domain ontology that supports decision-making is granted and so is the existence of a database supporting the patients' information storage. In order to exploit both the advantages of the existing database and the developed domain ontology we came to develop a mapping mechanism between them, migrating database instances into ontological instances (individuals) (also called ontology population), by a query driven process of transforming the database instances that are the response to a given query.
Bearing in mind the concepts of personalized medicine and evidence-based medicine, section II presents the AI-CARE monitoring system by discussing important issues of the ontological-based CDSS development, and the architecture of the patient-centric EHR (subsection II.A and II.B, respectively), and by focusing on the alignment among Ontology and EHR entities (subsection II.C), as well as test scenarios and validation (subsection II.D). Conclusions and issues for future research are discussed In Section III.
AI-CARE MONITORING SYSTEM
FOR PATIENTS WITH BDI
The AI-CARE monitoring system aims to provide an effective monitoring for patients with bipolar disorder.
Bipolar disorder (BD), also known as manic-depressive illness, is a severe mental illness, thought to be caused by an interaction of genetic and environmental factors. BD which is triggered by stressful life events, is often misdiagnosed and/or not sufficiently treated, and is associated with a high risk of suicide. Obviously, considering the important aspects of BD, such as the early onset, natural history, lifetime prevalence (1 to 5 % in general population, estimated in different studies), mental anguish, high rate of recurrence (>90 % of patients
who have a single manic episode will have future episodes), and psychiatric/medical morbidity, justify the need to develop an intelligent system in order to longitudinally monitor the evolution of this complex and heterogeneous disease in bipolar patients [17].
The AI-CARE monitoring system utilizes the ontological-based CDSS as part of an electronic health record in order to be beneficial to the bipolar patients, aiming to provide «the right patient with the right drug at the right dose at the right time» and tailoring the medical treatment to the individual characteristics, needs and preferences of a bipolar patient during all stages of care (diagnosis, treatment and follow-up, prevention). Also, is capable of identifying the characteristics of patient-subpopula-tions that do not benefit from the recommended therapy leading to new expert knowledge, new research and prospectively to new recommendations [12].
DEVELOPMENT OF AN ONTOLOGICAL-BASED
CDSS
In order to develop electronic support for clinicians and health care professionals, a knowledge-based system is needed to be developed consisting of a knowledge-base that represents facts about the disorder and an inference engine that can reason about those facts and use rules and other forms of logic to deduce new facts or highlight inconsistencies.
Ontology
The most prominent language for implementing ontologies is Web Ontology Language (OWL). The basic structure of OWL are classes, properties and individuals, which are members of classes. OWL properties are binary relationships and are distinguished in object properties (relate two individuals) and datatype properties (relate an individual with a literal value). Also OWL can define hierarchies of classes and properties, property domain and range restrictions, value restrictions, cardinality, existential and universal quantification restrictions on the individuals of a specific class. Base of OWL is Description Logics (DL) [1].
Ontologies can be distinguished by the subject of the conceptualization, such us knowledge representation, upperlevel, domain and application ontologies [7].
During the definition of a medical domain, achieving formalization of the domain terminology and categorization, is a desired result. In this attempt, formal ontologies such as SNOMED CT [9] or other formal approaches, offer great advantages in formal rigor and inference power. Nonetheless, they limit the expressiveness of the domain representation and design to an upper level description
[8, 15]. Considering these limitations, our attempt to define the bipolar disorder domain integrates: (i) a vocabulary of terms along with concept definition and their inner-relationships, which is offered by formal ontologies, and (ii) a more specialized description that is geared around the concepts related to the patient condition monitoring evolving in time, as presented in Fig. 1.
In order to describe the changing aspects of the disease in terms of states, state transitions and processes, the ontology needs to be dynamic. In our implementation, we design the initial static ontology, describing the main concepts of Bipolar Disorder, using the Protégé ontology editor1. The static ontology is converted into dynamic using the CHRONOS plugin of Protégé [13]. The main concepts describing the domain of Bipolar Disorder are distinguished into dynamic entities (entities which evolve in time) and static entities (entities which do not evolve in time).
Ontology is populated with real data collected from 10 patients whose condition is monitored over a period of a few days to a year.
Rules
We derive new knowledge from the assertions in ontology adopting Semantic Web Rule Language rules (SWRL)2, which is the most prominent lan-
guage for editing such rule. A SWRL rule presents an implication between an antecedent and a consequent so that the intended meaning is: whenever the condition specified in the antecedent hold, then the conditions specified in the consequent must also hold.
An example of a treatment recommendation rule is presented, resulting into the suggestion of medical treatment. It evaluates the medication the patient is receiving and the type of symptoms the patient presents.
In the case that the patient is first diagnosed, receiving no medication and the symptoms suggest existence of a manic episode then the rule directly suggests medical treatment (Lithium, Li; Valproate, VPA; atypical antipsychotic, AAP). Necessary information for the rule is included in the classes Personal Health Record (PHR), PatientState, Episode, Therapy, and Medicine.
The rule is expressed in DLs [1] as: PHRO( 3 PatientState.state = inEpisode)n (3 Episode.type = manic) n Therapy n^ 3 Medicine ^Recommendation (Start Therapy with Li/VPA/AAP or combination of two medicines) The rule is expressed in SWRL as: PHR(?phr), MedicalHistory(?medHist), Recommendation(?rec),
includesInitialEvaluation(?phr,?ev_medHistory),
1 http://protege.stanford.edu/ 2 http://www.w3.org/Submission/SWRL/
includesInitialEvaluation
(?ev_medHistory,?medHist), Event(?ev_medHist),
Interval(?int_medHist),
during(?ev_medHist,int_medHist),
evaluationOP(?phr,?ev_eval),evaluation
(?ev_eval,?eval),equal(?eval,»mild to severe»),
Event(?ev_eval),Interval(?int_eval),
during(?ev_eval,?int_eval),interval
Before(?int_medHist,?int_eval),
text(?rec,?txt), equal(?txt,
«Recommend Li/VPA/AAP or 2 drug
combination») ^ recommendationBelongs
(?rec,?phr)
DOCUMENTED CLINICAL KNOWLEDGE
FOR BDI
Clinical practice guidelines refer to «systemati-cally developed statements to assist practitioner and patient decisions about appropriate health care for specific clinical circumstances» according to the definition of Institute of Medicine (IOM), and are mostly relied on comparative clinical trials [19]. They are commonly utilized by the evidence-based medicine in making clinical decisions to improve the care process [4]. The scope of medication guidelines is to aid the interaction of clinicians and patients in developing the most effective treatment strategy minimizing the side effects.
Clinical guidelines for Bipolar I Disorder (BDI)
We selected evidence-based clinical practice guidelines (e.g. Australian and New Zealand, British Association for Psychopharmacology) related to different aspects of care for BDI, as well as other systematic reviews for BD [5]. Such guidelines utilize evidence-based knowledge for treating and managed patients with BDI defined by specific clinical criteria.
User Scenarios for Bipolar I Disorder
Bipolar Disorder is usually classified within the context of the Diagnostic and Statistical Manual of Mental Disorders (DSM), which differentiates between bipolar I disorder, bipolar II disorder, bipolar disorder not otherwise specified (Bipolar NOS), and cyclothymia.
Presently, we provide user diagnostic and treatment scenarios for the clinical presentation of BDI (at least one manic or mixed episode). Diagnostic scenarios consider specific information of screening and assessment tools and persons' medical/family and past psychiatric history. Diagnosis of bipolar I disorder follows the established criteria of the DSM-IVor DSM-V for a manic or depressive episode along with their severity, considers the psychi-
atric or/and medical comorbidities and hinders misdiagnosis, especially with unipolar disorder (major depressive disorder).
Based on the aforementioned evidence-based guidelines [5], the diagnostic scenarios are designed to guide the clinicians during the diagnosis procedure (mental state examination, initial evaluation considering the differential diagnosis, assiduous psychiatric examination addressing the different patterns of BD emergence, unobserved comorbidities and related disorders) and the diagnostic accuracy when patients fail to respond to treatment [5]. Also, we developed the user scenarios for BDI treatment options following the dynamic disease course.
The ontology-based CDS system provides diagnosis and treatment recommendations related to the patients' mental state (acute episodes, euthymia), alerts related to crucial mood swings and medication noncompliance, preventive care reminders about monitoring procedures (e.g. extrapyramidal symptoms, lithium serum levels, weight gain, diabetes screening, hyperlipidemia assessment), and warnings related to changes in symptom complex, as well as assists clinicians with decision-making and in developing a personalized disease management.
Longitudinal monitoring of bipolar patients is performed by the developed system to evaluate the presence or absence of symptoms, psychiatric and/or mental comorbidities, medication adherence, and to identify therapeutic drug safety and tolerance. Warnings have been placed in decision nodes with relevant annotation from the literature in order to yield the appropriate hints and alerts to the clinicians on realtime and at the time of care. The monitoring functionality can be further enriched by means of inputs received from biosensors (i.e. biosignals) and smart-phone applications (e.g. voice analysis) accompanied by inputs (paper-based and electronic-based data like life charting) from the user's environment (family, carers) or the user himself.
The ontology-based CDS system is highlighted as a crucial component of the patient-centric EHRs for BD. It is implemented using a networked EHR platform, whether the knowledge is available from a repository outside the local site and is accessed, but not incorporated into the local EHR.
ARCHITECTURE OF A PATIENT-CENTRIC EHR
Consistent with the conceptual view of longitudinal monitoring of patients with BD, we created an Electronic Health Record (EHR) for bipolar patients. It encloses five of the essential components of the EHRs: (i) administrative processes, (ii) health information and data, (iii) communication and connectivity, (iv) results management, and (v) decision support [16].
Presentation Tier
Application Tier
Client
Rich Web Interface
Application Server
Business Logic
Business Objects
Security Framework
RBAC Mechanism
Encryption/Decr yption Mechanism
Data Access
Data Access Object Relational
Object Mapping
Intelligence Tier
Ontology Server
Ontology
Rules
Data Tier
Database Server
Data Access
— —
— —
Database
-----
Fig. 2. Platform architecture - tiers
The ontology-based CDS system delivered with the EHR as depicted in the platform architecture of Fig. 2 will provide clinicians with suitable tools achieving a better day-to-day clinical decision making.
The architecture of the integrated platform consists of four main tiers (Fig. 2) namely, a) Presentation tier, b) Application tier, c) Intelligence tier, and d) Data tier.
Presentation tier
The first tier, called presentation tier, is the top layer of the integrated platform. This layer is responsible for exchanging information between the general stakeholders and the system. Its main focus is to provide advanced usability and visualization functionality along with a simple and rich graphical
Л Г, № Я HI' CM twV ■», û pYl <1 l'Ui
/ S /■ / / / / / / / f
f S 4 « J S S ¿ J У 4?
"■ГГ-К,
IpVfH *«nf"K" r-J >Ч«Ч
f f f t <t f ¿ f f s * f
y f -î ? f ■* Í® f' í» J
Fig. 3. Graph presenting patient's episodes with respect to the substance administration over time СИБИРСКИЙ НАУЧНЫЙ МЕДИЦИНСКИЙ ЖУРНАЛ, ТОМ 36, № 1, 2016 91
user interface (GUI), in order to present the stored information to the end users. Since it is responsible for the interactions between users and the system, it provides access to the EHR through the web browser and among other useful graphical user interface components it makes statistical graphs available for use for intuitive visualization of the current status of patients in terms of episodes and substance administration over time (Fig. 3).
Application tier
In the middle of the tier platform, there is the application tier. This is in charge of the control of the system's operations, achieved by performing detailed processes. It consists of three sub-tiers: (i) Business logic, (ii) Security and (iii) Data access.
The operations of the business logic sub-tier concern the processing of a heterogeneous information data set. Some of the core functionalities that stand out are the support in retrieving patients' information; exporting patients' data into XLS format file; the dynamic clinical forms generation; the retrieval and process of data for the visualization of diagrams; the comparison of patient's re-examination data and printing capabilities. Furthermore, the business logic sub-tier is responsible for providing a robust working environment, coping with errors during execution and continuing the system's operation despite the potential input inconsistences. That is done mainly by preventing users from entering erroneous input, guiding them towards its proper use and performing in a satisfactory way, such that it does not intercept user's flexibility, agility and operability.
The information to be managed by this tier is patients' clinical and demographic data, users' personal data and access credentials, user's roles, users' access to patients, informed consent documents files, substance administration data and information that concern the dynamic data entry clinical forms. The functionality in this section addresses consistent terminologies/ vocabularies/ standardized transactions, data correctness, and interoperability (i.e. ICD-10, NOM).
The significance of security for the integrated platform is vital since it manages sensitive data when at the same time it is accessed by a variety of users which as stakeholders they have different roles and responsibilities according to their position and skills. The heterogeneity that characterizes those users raises the necessity of utilizing a Role Based Access Control (RBAC) [6] mechanism to regulate user actions within the system. These roles can guarantee that no user can perform ineligible acts.
Moreover, encryption/ decryption mechanisms are used to further secure users' passwords and pa-
tient's sensitive data. Data stored in an encrypted way can ensure the confidentiality in case that a third party breaches the database access. Users' credentials are stored in an encrypted unidirectional way. That means that passwords are not decryptable and thus cannot be recovered. On the other hand, all patients' sensitive data (data that can be used for identifying the patient) are stored in an encrypted bidirectional way within the database, allowing them to be decrypted, since the identification of the patient is required.
Finally the Secure Socket Layer (SSL) [3] protocol is utilized in order to preserve the secure data exchanged through a public-and-private key encryption mechanism which includes the use of a digital certificate.
The data access sub-tier handles all the logic regarding data storage and management. That is achieved by providing an abstract interface by using Data Access Objects (DAO) and thus delivering specific data operations without exposing details of the database. Finally, data persistence is achieved by adopting the Object Relational Mapping (ORM), which solves object-relational impedance mismatch problems by replacing direct persistence-related database accesses with high-level object handling functions.
Intelligence tier
The intelligence tier basically corresponds to the knowledge-based system, described in section A, offering knowledge extraction from the existing patient information and concluding into clinical decision support through treatment recommendations and alerts.
Data tier
The bottom tier is the data tier, which constitutes the database server of the integrated platform to store all the information data. In that way, data is kept neutral and independent from the rest of the tiers offering improved scalability and performance. The design of the relational database model was based on HL7-RIM3, which is the cornerstone of the HL7 Version 3 development process and provides to the database a flexible and extensible structure.
ALIGNMENT OF THE ONTOLOGICAL-BASED
CDSS AND THE PATIENT-CENTRIC EHR
The main contribution of this paper is the assignment of the ontology entities to the EHR entities, in order to enable communication and connec-
3 http ://www.hl7 .org/implement/standards/rim. cfm
Table 1
Relations between ontology and ehr entities
Ontology Classes/Subclasses EHR Task Categories/Subcategories
Dynamic Entities
PatientState: patient's current state (in euthymia or in an episode) Clinical Data - Clinical Picture Diagnosis Patient's Diagram
Symptom: the symptom (type, severity) Clinical Data - Clinical Picture
Therapy: the therapeutic approaches a patient may receive (medication, hospitalization, psychotherapy) Clinical Data - Biological Therapy - Psycotherapy - Psychoeducation
Medicine: the substance administration of the patient Clinical Data - Biological Therapy - Drug Information Medicine Patient's Diagram
Monitoring: (i) FunctionTest the tests a patient is submitted to (imaging tests, laboratory tests etc.) (ii) Biosignal keeps the information of biosignals recorded from sensors applied on patient Clinical Data - Laboratory Testing Process - Longitudinal Monitoring
AssessmentTools (i) Interview (ii) Questionnaire (iii) RatingScale (iv)EvaluationTest Clinical Data - Clinical Picture - Longitudinal Monitoring - Psychometric Approach Diagnosis
Static Entities
Patient: patient's personal and demographic information General Information Contact
Episode: the type (manic or depressive) and severity of an episode Clinical Data - Clinical Picture Patient's Diagnosis Patient's Diagram
Diagnosis: the type of the disorder (Type I or Type II) and whether the patient suffers from rapid cycling Diagnosis (according to DSM- IV or DSM- V criteria for bipolar disorder) Clinical Data - Past Psychiatric History Patient's Diagram
PatientHistory: patient's medical and past psychiatric history (age of onset, heredity, number of manic or depressive episodes, previous medication) Clinical Data - Personal History - Medical History - Past Psychiatric History
FamilyHistory: patient's family history Clinical Data - Family History
SideEffect: possible side effects of a medicine Clinical Data - Drug Information - Longitudinal Monitoring
DifferentialDiagnosis: differential diagnosis for the diagnosis procedure. It is a combination of: (i) ClinicalEvaluation (ii) SubstanceEvaluation (iii) DepressionCriteria Clinical Data - Clinical Picture - Psychometric Approach
MedicalCause: whether the clinical evaluation suggests other causes than bipolar disorder Clinical Data - Clinical Picture - Psychometric Approach
a
JDBC API
Database
Constraints
DB Model
Mapping Algorithm
Fig. 4. Mapping process
OWL Domain Ontology
SQL tables
Table 2
Table Name Attributes
SubstanceAdministration id, dosequantity, frequency, adm_r, startDate, endDate, actId, substanceAdministrationCodeId[F K]
ActInstance id, type, actId[F K], observation^ K], reviewed, substanceAdministrationId[F K], diagnosisId[F K], consentFieldId[F K], signalFieldId[F K]
Participation id, actClass, actMood, code, title, desc, statusCode, roleInstanceId[F K], actInstanceId[F K]
RoleInstance id, statusCode,validForm, validUntil, personId[F K], roleCodeId[F K]
Person id, classCode, username, password, enabled
tivity among the ontological-based CDSS and the patient-centric EHR [18] (Table 1).
The EHR records longitudinally the patient health information, including demographics, contact information, clinical data (e.g. patient's personal, medical, and past psychiatric history, mental state examination, laboratory and data, and electronic diary mood reports, drug information), diagnosis, patient's diagram (visualization of the current and/or previous status of patients in terms of episodes/ disease state and drug administration over time), as well as an up-to-date drug database. The EHR supports indirectly the developed ontology-based DCSS leading to advanced health care quality.
Moreover, in order to achieve the integration between the ontology and the database, a mapping mechanism able to define the correspondences between the entities of the database and the ontology schema, is needed. Although relational databases are based on closed-world assumption whilst ontologies use open-world semantic, at a conceptual level, a database and an ontology are semantically related and correspondences are established between the database components and the ontology components. For instance, an attribute in a relational database schema may correspond to a property in an OWL ontology.
In our approach, a naive mapping procedure is adopted retrieving data from SQL queries, through JDBC4, applied over the source database and refor-
4 http ://www.oracle.com/technetwork/j ava/j avase/
jdbc/index.html
mulates the results, in terms of the target ontology, through OWL API5. Such mapping specifies the ontology population from the data in the database.
Fig. 4 depicts a simplified conceptual view of the mapping process. As a simplified model, it does not show the complex nature of its components (e.g. database, tables, or ontology classes), rather the interaction between DB Model and Ontology Model, which is feasible through a mapping procedure with OWL API.
A query example retrieving the medicine information for a specific patient is presented. The database tables that relate with each other are the following, see Table 2. The sql query is presented in Table 3.
TEST AND VALIDATION OF THE AI-CARE
MONITORING SYSTEM
The clinical use of the AI-CARE system is expected to provide answers to relevant questions related to the individualization of diagnosis, treatment approaches and effectiveness of treatment, transition hazard from major depressive episodes to manic, and malignant types of BD.
More specifically the monitoring system will be tested under the following issues in BDI regarding the knowledge residing in the ontology model and the patient information in the EHR model: (1) Diagnostic criteria of BD in terms of early and complete
5 http://owlapi.sourceforge.net/
recognition of disease, (2) Evaluation of clinical effectiveness in terms of timely, and regular treatment, and acceptability of treatment dosage, (3) Evaluation of current patient's condition, (4) Identification of directly transition from depressive episodes to manic, and vice versa (5) Evaluation of treatment effectiveness in terms of quality of life, and improvement in symptoms, and (6) Identification of high risk patients (treatment non-response in BDI patients).
CONCLUSIONS
In our study, we align an ontological-based CDSS and a patient-centric EHR providing an on line monitoring tool, which seek to support psychiatrists and mental health professionals in tailor evidence-based practice in day-to-day clinical decision making for the longitudinal monitoring of each bipolar patient. Apart from the test and validation, the ontology population is an ongoing process. In the future, we aim to adjust the monitoring system for other types of BD and for epilepsy. The presented AI-CARE ontological-based monitoring system combines both personalized and evidence-based medicine in order to promote the care process for the patients suffering from bipolar I disorder.
ACKNOWLEDGEMENTS
This work is partially supported by the «AI-CARE» project of the «COOPERATION 2011» framework under the NSRF 2007-2013 Program of the Greek Ministry of Education, Lifelong Learning and Religious Affairs.
REFERENCES
1. Hristoskova A., Sakkalis V., Zacharioudakis G., Tsiknakis M., De Turck F. Ontology-Driven Monitoring of Patient's Vital Signs Enabling Personalized Medical Detection and Alert // Sensors. Vol. 14, N 1. P. 1598-1628, 2014.
2. http://www.fda.gov/downloads/ScienceResearc h/SpecialTopics/PersonalizedMedicine/UCM372421.pdf
3. Carmeli B., Casali P., Goldbraich A., Goldste-en A., Kent C., Licitra L. et al. Evicase: an evidence-based case structuring approach for personalized healthcare // Stud. Health Technol. Inform. 2012. 180. 604-608,
4. Sackett D.L., Rosenberg W.M., Gray J.A., Hay-nes R.B., Richardson W.S. Evidence-based medicine:
what it is and what it isn't,» [editorial] // BMJ. 1996. 312. 7023. 71-72,
5. https ://healthit. ahrq.gov/sites/default/files/docs/ page/09-0069-EF_1.pdf
6. Maedche A., Staab S. Ontology learning for the Semantic Web // IEEE Intelligent Systems. 2001. 16. 2. 72-79.
7. Thase M.E. Bipolar Depression: Issues in diagnosis and treatment // Harvard Rev. Psychiatry. 2005. 13. 257-271.
8. Baader F., McGuinness D.L., Patel-Schnei-der P.F., and Nardi D. The Description Logic Handbook: Theory, Implementation, and Applications, Cambridge University Press, 2nd ed., 2007.
9. Gomez-Perez A., Corcho-Garcia O., Fernandez-Lopez M. Ontological Engineering, Springer-Verlag New York, Inc., 1st ed., 2003.
10. Heja G., Surja G., Varga P. Ontological analysis of SNOMED CT // BMC Med. Inform. Decis. Mak. 2008. 8. S8.
11. Schulz S., Stenzhorn H., Boeker M., Smith B. Strengths and limitations of formal ontologies in the biomedical domain // Rev. Electron Comun. Inf. Inov. Saude. 2009. 3. 1. 31-45.
12. Halland K., Britz K., Gerber A. Investigations into the use of SNOMED CT to enhance an OpenMRS health information system // South African Computer Journal. 2011. 47. 33-46.
13. Preventis A., Petrakis E., Batsakis S. Chronos Ed: A tool for handling temporal ontologies in Protege // (IJAIT), 2014.
14. Timmermans S., Mauck A. The promises and pitfalls of evidence-based medicine // Health Aff (Millwood). 2005. 24. 1. 18-28.
15. Connolly K.R., Thase M.E. The Clinical Management of Bipolar Disorder: A Review of Evidence-Based Guidelines // Prim. Care. Companion CNS Disord. 2011. 13. 4. PCC.10r01097.
16. Seckman C. Electronic health records and applications for managing patient care // In: Elsevier and typesetter Toppan Bes. 2013. 87-105.
17. Ferraiolo D., Cugini J., Kuhn D.R. Role-based access control (RBAC): Features and motivations // 11th ACSAC, 1995.
18. BhiogadeM.S. Secure socket layer // Computer Science and Information Technology Education Conference, 2002.
19. Thermolia C., Bei E.S., Petrakis E.G.M., Kri-tsotakis V., Tsiknakis M., Sakkalis V. Designing A Patient Monitoring System for Bipolar Disorder Using Semantic Web Technologies // In Proc. IEEE EMBC, Milano. Italy, 2015.
СИСТЕМА МОНИТОРИНГА ПАЦИЕНТОВ С БИПОЛЯРНЫМ АФФЕКТИВНЫМ РАССТРОЙСТВОМ I ТИПА НА ОСНОВЕ ОНТОЛОГИЧЕСКОГО ПОДХОДА
Хриса Х. ТЕРМОЛИА1, Екатерини С. БЕИ1, Эврипидес Г. М. ПЕТРАКИС1, Вангелис КРИТСОТАКИС2, Вангелис САККАЛИС2
1 Школа электронной и вычислительной техники, Критский технический университет 73100, Греция, г. Ханья, кампус Акротири
2Институт компьютерных наук, Фонд исследований и технологий 71110, Греция, г. Гераклион, Н. Пластира, 100
Цель исследования - обеспечение системы мониторинга пациентов, объединяющей систему поддержки принятия клинических решений (CDSS) и электронных медицинских карт (EHR) и помогающей психиатрам и врачам первичного звена в обеспечении существующих потребностей здравоохранения в области психических заболеваний, связанных с лечением и регулирование биполярным аффективным расстройством I типа (BDI). Предложенная система мониторинга состоит из системы EHR, основанной на эталонной информационной модели медицинского стандарта 7 уровня (HL7-RIM) и онтологической CDSS, использующей возможности семантической паутины. Система мониторинга разработана на основании руководств по доказательной медицине и медицинских карт пациентов и позволяет кодировать и обрабатывать эту информацию для последующего назначения рекомендаций выбора и оповещения клиницистов с целью улучшения оказания психиатрической помощи. Учитывая данные соответствующих клинических руководств, а также истории болезни пациента, система мониторинга может способствовать принятию персонализированных решений при длительно текущем BDI. В качестве онлайн-инструмента мониторинга мы предлагаем систему AI-CARE, которая может оказать большую помощь в клинической практике.
Ключевые слова: система поддержки принятия клинических решений, электронные медицинские карты, семантическая паутина, онтология, биполярное аффективное расстройство.
Термолиа Х.Х. - студент, e-mail: cthermolia@ intelligence.tuc.gr
Беи Е.С. - кандидат технических наук, докторант, сотрудник лаборатории, e-mail: abei@isc.tuc.gr Петракис Э.Г.М. - проф., зав. лабораторией , e-mail: petrakis@intelligence.tuc.gr Критсотакис В. - инженер, e-mail: vkrits@ics.forth.gr
Саккалис В. - главный научный сотрудник лаборатории, e-mail: sakkalis@ics.forth.gr
