General Terms:

• Concepts - Raw-Data concepts represent medical raw data such as measured parameters, e.g. the raw-data concept "Systolic blood pressure". Abstract concepts represent the Interpretation and summarization of raw-data, e.g. the abstract concept "High Blood Pressure" represents the value of the integrated interpretation of the raw-data concepts "Systolic Blood Pressure" and "Diastolic Blood Pressure".

• Temporal patterns - abstract concepts that represent the fact that certain value and temporal relationship constraints hold among several raw and/or abstract concepts (e.g., “two weeks of moderate anemia followed by two to three weeks of low blood pressure”).

• Declarative knowledge - definitions and properties of raw and abstract concepts. For example, the definition of “High Blood Pressure during pregnancy” or the allowed values for interpretations of blood glucose. Knowledge about abstract concepts typically requires a context, such as “pregnancy”, age or gender.

• Declarative [medical] domain knowledge base – a set of concepts from a particular medical domain (e.g., Cardiology), their properties, and their inter-relations (e.g., Severe-anemia is derived from Hemoglobin).

• Temporal abstraction - generation of interpretations (abstract concepts) from raw time-oriented data, based on the domain’s declarative temporal-abstraction knowledge base.

• Temporal abstraction ontology- a pre-defined high level, generic ontology for structuring all domain-specific declarative knowledge bases (analogous to a database scheme). It includes high-level concepts such as “raw-data concepts”, a subset of which, in a medical domain, are “laboratory measurements”; or “interventions”, a subset of which are “medications”, or “temporal patterns”, which include linear and repeating patterns.

• Procedural knowledge – representation of a therapeutic or diagnostic process, such as the decomposition of a guideline into plans and sub-plans, in sequence or in parallel, implying a particular procedural workflow. For example, the plan for "gestational diabetes management" might be decomposed into “blood glucose monitoring” and “dietary intervention”, both of which should be performed in parallel.

• Clinical guidelines – recommended evidence-based procedures for management of certain diagnostic and/or therapeutic situations, such as a particular disorder (e.g., Type II diabetes).  Typically released as text documents by professional medical associations or governmental bodies (e.g., National Institute for Health and Clinical Excellence (NHS in England) or the New Zealand Guidelines Group). Clinical guidelines include both the procedural knowledge describing the treatment process and the declarative knowledge describing relevant concepts, such as “moderate anemia” in the guideline’s context.

• Knowledge base- a unified repository that stores declarative and procedural knowledge for a particular domain, e.g. “bone marrow transplantation.”

• Knowledge server - one or more knowledge bases; supports storage and retrieval of knowledge instances.

Our Tools:

• Knowledge specification tools –an integration of graphical tools, knowledge bases, knowledge servers, and methodologies for acquisition and maintenance of both procedural and declarative knowledge; typically used by a medical expert and/or a knowledge engineer. Structuring the text is called “mark up.”

• Digital guideline library - a set of graphical tools for guideline classification by controlled medical vocabularies, semantic markup, storage, context-sensitive search (for terms within a particular context, such as “eligibility condition”), concept-based search (based on the guidelines’ classification), retrieval, browsing.

• Temporal abstraction services - computational services for interpretation of time-oriented patient data, using the domain’s declarative temporal-abstraction knowledge.

•  Intelligent temporal mediator – a service that integrates the raw time-oriented data concepts with relevant declarative temporal-abstraction knowledge to support temporal-abstraction tasks such as query, monitoring, interpretation, and summarization of time-oriented patient data.

• Guideline application services - graphical tools and computational services for applying the guideline to a particular patient, using the digital guideline library and the intelligent temporal mediator.

• Temporal information visualization tools – interactive graphical tools for visualization and exploration of individual and multiple patients' time-oriented raw and abstract concepts (using the intelligent temporal mediator), as well as interactive investigation and analysis of large numbers of longitudinal clinical records, such as for display of associations among concepts over time.

• Mapping the knowledge to a local database – A process that uses graphical tools and controlled medical vocabularies for mapping a particular domain knowledge base into a local database. E.g., specifies what local term stands for “fasting blood glucose.”

Our Platform:

Figure 1. Our platform for the automation of guideline-based care and for intelligent query, interpretation, monitoring and analysis of longitudinal patient data. It is an overall view of the architecture and main components.

1. Knowledge Library and Knowledge Specification tools:
   This module includes: (a) a digital library and tools for specification, maintenance, storage and retrieval of the procedural knowledge of guidelines; (b) tools for specification, maintenance and storage of declarative knowledge. The tools enable medical experts and knowledge engineers to collaborate in the specification and maintenance of both types of knowledge, a necessary prerequisite for support of guideline-based care.

2.  Tools for mapping knowledge to local CPRs:
   This module includes an “adapter” from the declarative knowledge base to the medical databases. To be general, terms in knowledge bases are from controlled medical vocabularies that are part of the UMLS.

3. Clinical Decision Support System and Services:
   These include tools for (1) runtime customized application of clinical guidelines based on the particular patient’s data and on the guideline’s procedural knowledge; and for (2) intelligent query, monitoring, exploration, and analysis of time-oriented clinical data for individual patients (e.g., by care providers) as well as for large groups of patients (e.g., by health-care managers), using the domain’s declarative temporal-abstraction knowledge. Thus, we support guideline application, quality assessment, and decision making.