Adaptable Personalized Care Planning via a Semantic Web Framework

of 6

Please download to get full document.

View again

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
6 pages
0 downs
Adaptable Personalized Care Planning via a Semantic Web Framework
  Adaptable Personalized Care Planning via a Semantic Web Framework Syed SR ABIDI a,  and Helen CHEN  b   a  Health Informatics Research Group, Faculty of Computer Science,  Dalhousie University, Halifax, Canada b  Agfa Healthcare, Waterloo, Ontario, Canada Abstract. This paper presents a longitudinal patient care planning system that automatically and pro-actively generates adaptive patient-specific healthcare  plans . The objective is to assist medical practitioners to determine the best discourse of clinical care based on (a) the patient’s current medical profile, (b) latest medical knowledge (c) institution-specific clinical pathways, and (d)  personalized healthcare educational programs. We present a novel semantic web framework that allows for the synthesis of heterogeneous operational and medical information and knowledge resources, and renders the technical basis for a services-oriented architecture to generate and orchestrate patient-specific healthcare plans. Keywords: Medical Knowledge Management, Clinical Pathways, Care Planning, Semantic Web, Web Services 1.   Introduction In Canada, and likewise around the world, there is a growing realization that health human resource planning and evidence-based practices are vital to deliver effective healthcare services. Currently, clinical care is discharged in a reactive and sub-optimal manner without due consideration to optimization of available healthcare resources and the proactive planning of the care regime. Despite recent developments in health information collection, sharing and processing infrastructures, there is a realization that the health informatics research agenda should include the development of new methods, leveraging emerging knowledge-mediated technologies, for optimal clinical care planning and informed decision-making as per the individual patient’s profile and healthcare site’s operational requirements. Lifelong patient care management, in its entirety, is a complex process that can be investigated from a variety of perspectives. Our belief is that the systematic application of advance knowledge management approaches—in particular the emerging semantic web framework—can lead to the development of a knowledge-mediated patient care  planning systems that automatically and pro-actively generate adaptive patient-specific healthcare plans that may guide the long-term clinical, therapeutic and rehabilitation care process for individual patients within a specific healthcare setting. Our approach is to investigate proactive patient care management, whereby a  patient-specific care plan accounts and satisfies the evolving healthcare needs of a  patient in the continuum of care. Functionally speaking, for each individual patient a  care plan is to be conceived through a systematic interplay between (a) patient information sourced from online health reporting documents; (b) best evidence manifested in clinical practice guidelines, (c) clinical pathways stipulating the care  process, resource constraints and therapeutic costs; and (d) a (semantic) web of heterogeneous medical knowledge resources. The systematic synthesis of these determinants for patient care realizes a process-oriented and knowledge-mediated solution to patient-centric healthcare—the solution is CarePlan . In this paper, we present a semantic web based framework [1] [2] for patient care  planning. The featured CarePlan framework generates adaptive patient-specific CarePlan  depicting a roadmap of prospective events, actions, schedules, (site specific) resource implications and expected outcomes. Here, we discuss the functional components of the CarePlan framework. 2.   CarePlan: A Conceptual Overview A patient-specific CarePlan can be envisaged as a personalized clinical pathway that encapsulates: (a) a chartered discourse of clinical care activities—mandated by clinical guidelines/knowledge and validated by intelligent proof engines—to address a patient’s current healthcare needs; (b) a recorder of the temporal sequence of medical events, actions and outcomes as they occur in the longitudinal continuum of care; (c) the gateway to source case-specific evidential and experiential medical knowledge, for  both practitioners and a computer system, in order to both reason about the patient’s diagnosis, prognosis and to formulate the patient’s therapy and rehabilitation; and (d) the basis for generating patient-specific educational interventions. CarePlan  can therefore be regarded as a rich temporal, process-centric, patient-specific abstraction of the ever-evolving dynamics of total patient care management in a specific healthcare setting. In practice, when a patient with a chronic health problem enters the healthcare system, his/her personalized CarePlan will be generated based on his/her current health  profile; and as the patient evolves his/her CarePlan will dynamically adapt to meet the  patient’s current conditions, new medical knowledge, physician’s inputs and institutional workflows. The prime requirement for the CarePlan framework is to access, integrate, adapt and manipulate heterogeneous healthcare knowledge in response to available patient information (as shown in Figure 1). The emerging  semantic web framework  , in the realm of knowledge management, provides both the theoretical basis and the applied methods to (a) represent knowledge in ontology guided formalisms; (b) reason over the knowledge using proof engines; (c) morph the different knowledge modalities to form a unified knowledge object; (d) adapt standard care plans towards personalized healthcare plans by reasoning and morphing the available knowledge; (e) integrate web-services for the composition of healthcare plans; and (f) establish trust over the  personalized CarePlan by way of validation through knowledge-based proofs. The abovementioned CarePlan  objectives demand a sophisticated info-structure that features the following ten core functionalities: a)   Ability to collect patient information from multiple sources and in different modalities to create a patient information model.  b)   Ability to abstract patient information from the information model to yield an episodic profile of the patient, which would be the basis for selecting the relevant clinical practice guideline(s).  c)   Semi-automatic methods to transform paper-based Clinical Practice Guidelines (CPG) and also clinical pathways to standard representational formalisms. d)   Functional links between the patient information model and the CPG model to ensure that patient information can be seamlessly input to a computerized CPG to chart evidence-driven actions. e)   Functional and conceptual links between CPG and the actionable clinical  pathways to generate a CarePlan  which is based on best clinical evidence and congruent with institutional resources. f)   A semantic web of best medical knowledge serving as the knowledge backbone  responsible for maintaining currency, quality and validity of the CarePlan  solution. g)   An interface to the semantic web that allows to (a) seek both evidential and experiential knowledge pertaining to the medical problem at hand; (b) morph the different knowledge modalities to realize a seamless holistic knowledge object that can be used by the  CarePlan  info-structure. h)   A library of CPG objects characterized and indexed based on domain ontology. i)   A library of clinical pathway components characterized and indexed based on domain ontology and the institutional workflow.  j)   A web service orchestration framework that incorporates logical proof engines for creating valid web services—i.e. the personalized CarePlan.  The CarePlan  web services framework leverages on the semantic web of medical knowledge, the domain ontology, the institutional workflows and the proof engines. The output is a dynamically composed CarePlan  personalized to an individual’s needs and is compatible with the institutional workflow and is guided by the best current medical knowledge. 2.1.   CarePlan in action The full cycle of the CarePlan framework starts with (a) abstracting patient data from the information layer; (b) passing episodic patient information to the knowledge layer to select case-specific knowledge; (c) processing the knowledge, though the semantic web engine to generate a CarePlan at the planning layer; (d) validating the CarePlan  both by logic engines and practitioner input; and (e) passing the CarePlan to the medical practitioner. The CarePlan interface is a key element of the cycle as it provides  practitioners access to the semantic web of heterogeneous healthcare knowledge bases and distributed patient information. The CarePlan interface presents recommendations and pathways based on the patient condition and protocols. Most importantly, this interface accepts physician’s decisions and his/her resolutions in the cases of contradictory evidence or unusual circumstances. The role of the semantic web is central in generating a CarePlan. Given the medical knowledge, patient-specific facts and medical practice logic, the semantic web engine allows the inferring of new facts in the process of finding the best care plan for an active patient at the time of consultation (or point of care). The semantic web proof engines ensure that the actions prescribed in the patient’s care plan take into account all relevant information and are consistent with the best knowledge and medical practice. It also ensures that the outcomes generated from an action are taken into consideration as input to the next step in the CarePlan. The CarePlan is manifested in terms of adaptable workflows that are represented and executed as web services and are  activated when a new patient enters the system or any new information about an existing patient becomes available. Episode T1  Episode Tn Patient a Patient-SpecificDataHealth ReportingDocumentsPrimary-carePhysician/NurseEMR Based PatientData Episode - KnowledgeQueryKnowledge - EpisodeFeedbackPatientInformationTherapeuticInformationPatientInformationTherapeuticInformation    P  a   t   i  e  n   t   I  n   f  o .   M  o   d  a   l   i   t   i  e  s CarePlanLifelong Longitudinal Patient CareCare Plan Generation &Adaptation Web ServicesProof Engines KnowledgeCare Plan Plan Generator  AdaptedCare PlanMedicalPractitioners CarePlanDB    C  ar  eP l   anI  n t   er f   a c e (  H um an & DB )   Problem-SpecificDiscussions ClinicalDecision-SupportExpert’s TacitKnowledgeEducationalContent/ModulesBest Evidence(Literature) Semantic Web of HealthcareKnowledge PopulationHealth DataServiceEducationResearchMedicalPractitioners Clinical PracticeGuidelinesClinical Pathways    I  n   f  o  r  m  a   t   i  o  n   L  a  y  e  r   K  n  o  w   l  e   d  g  e   L  a  y  e  r   P   l  a  n  n   i  n  g   L  a  y  e  r   Figure 1: The functional diagram for CarePlan, illustrating the three functional layers 3.   CarePlan Functional Components The CarePlan technical solution features a confluence of technologies to support the description, discovery, composition, and operationalization of a patient-specific CarePlan in terms of automatically choreographed web services (see Figure 1). Five main functional components constitute the CarePlan framework. 3.1.    Patient health profile generation A comprehensive patient health profile is the starting point for the generation of a  patient-specific CarePlan. Patient health profile generation involves the collection, aggregation and representation of patient information srcinating from multiple EMR sources. The patient profile encompasses patient-defining measurements, operational, diagnostic and/or therapeutic details. We use the XML based Clinical Document Architecture (CDA) to formulate a high-level patient profile that will allow HL-7 linkages for information collection. For semantic interoperability between different information sources, we use the MESH medical terminological system.  3.2.    Healthcare knowledge semantic web Semantic Web is a logic-oriented framework for both representing and connecting heterogeneous knowledge resources [1]. Given that healthcare knowledge exists in a variety of modalities, ranging from tacit to experiential to explicit, we are working on the development of a healthcare semantic web  to interconnect the various healthcare knowledge resources. Our approach is to use CPG as an interface between the patient data and the healthcare semantic web because computerized CPG incorporate constructs to represent and manipulate both CDA based patient data and ontology- based medical knowledge. Our idea is to decompose CPG and Clinical Pathways (CP) into process-centric objects that entail decision logic, knowledge constructs and constraints that can be analyzed and validated by logic-based proof engines. A systematic collection of such knowledge objects will yield the CarePlan The healthcare knowledge semantic web will serve as the knowledge backbone  responsible for maintaining currency, quality and validity of healthcare knowledge used to develop CarePlans. We have developed a prototype ontology using Protégé [2] and a set of n3 rules. Having patient information in CDA format and the CPG content in RDF, we are able to use the domain ontology and rules to simulate the knowledge needs for specific clinical pathways using the Jena rule-based inferencing engine. 3.3.    Healthcare knowledge morphing The healthcare knowledge semantic web provides an interconnected encapsulation of the available healthcare knowledge. So, what is needed next is a mechanism for case-specific morphing (or fusion) of multi-faceted knowledge object detailing all available solutions, viewpoints and documented outcomes in response to the conditions reported in the patient’s profile. Our plan is to pursue knowledge morphing as a knowledge modeling activity that interpolates a knowledge link between two or more knowledge objects that share a discrete notion of contextual compatibility [3]. The knowledge link will allow for reasonable inferencing over the parent knowledge objects in order to establish the trust and comprehensiveness of the derived CarePlan. Our knowledge morphing approach is to (a) leverage a semantic web of knowledge resources, whereby the knowledge contents are conceptually identified and semantically annotated based on a global domain ontology; and (b) employ a view integration approach whereby the underlying conceptual schemas of the knowledge resources are integrated into the global schema of a CarePlan. 3.4.    Adaptive CarePlan generation Given a patient profile, this activity aims to generate a patient-specific CarePlan by dynamically adapting standard clinical pathways. We envisage the CarePlan as a specialized clinical pathway that is dynamically generated, through a semantic web  based web services framework, based on the patient information, the morphed knowledge and the semantically annotated clinical pathways. The semantic web proof engines will ensure the validity and applicability of the CarePlan vis-à-vis the patient  profile. The adaptive nature of the healthcare plan will ensure that it is proactively modulated to meet the changes in the patient profile, discovery of new healthcare knowledge, changes in institutional workflows or resources.
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks